Telling a different story

07. October 2026 · Write a comment · Categories: Actuarial, Cartoons, COVID-19, Economics, Education, Health, Prediction, Professions, Regulation, Risk

We shall not cease from exploration
And the end of all our exploring
Will be to arrive where we started
And know the place for the first time.

T.S. Eliot – Four Quartets

I feel like I am having a bit of a T.S. Eliot week. I have been writing about the OBR since the beginning of this blog in 2013, with their repeated failures as a forecaster forming the banner for the blog for many years.

What has brought me back to the OBR, as if for the first time, is recalling during the recent Funding the Future conference the words of the very good and wise friend of mine who was there with me. I had invited him to talk to one of my undergraduate groups at the University of Leicester during a professional skills session a few years ago. We were looking at past disciplinary cases of the Institute and Faculty of Actuaries and discussing what the actuaries involved could have done better.

In one of the case studies we considered, an actuary had produced a quarterly reserving report for a company, estimating the reserves on a ‘high’ basis of £74.5 million, a ‘low’ basis of £36.9 million and a ‘best’ basis of £52.1 million. The company then produced financial statements for the year, approved by the board of which he was a part, containing a reserve figure of £15.5 million. My friend told the students that the way to challenge in a situation like this is not to focus on the numbers, but instead see that each different number reflects a different story being told. So you need to look at what the story would need to be for the £15.5 million to be an appropriate figure and take issue with the things that don’t make sense within that story.

It was great advice then and still is, and it is definitely the way to look at the OBR I think. We find a story unsatisfactory if the characters are shallow, do not behave in a credible way, and the plot oversimplifies a much richer situation (something which every story must do to some extent to be manageable to the reader) in a deeply unsatisfying way.

Several other commentators have talked about the inability of the OBR to forecast anything (including me) – this is the equivalent of insufficiently fleshed out characters behaving in unbelievable ways. Richard Murphy, for instance, pointed out how ridiculous the proposed character plotline for inflation was at the last budget:

Whereas in fact of course the downward trajectory predicted at the last budget didn’t look remotely likely even then. The Consumer Prices Index (CPI) rose by 3.1% in the 12 months to August 2026, up from 2.9% the previous month.

However it is Dan Davies’ recent couple of Substack posts (one and two) which get to the heart of the problem in my view. Dan Davies’ contention is that it is the plot oversimplification which is the main problem here. As he says:

The system we have set up only seems to be able to communicate two states though – the signal is either “COMPLIANT – WAR CHEST” or “NONCOMPLIANT – BLACK HOLE”. It would be a significant improvement even for the OBR to add some measure of urgency or non-urgency to its press release – a sentence like “this surplus is well within the forecast error, however, so it should not be used to justify a change in policy”.

But, in his view, it is even worse than that:

In general, most things which a government spends money on will tend to have effects on the economy which go well beyond the five year forecast. The OBR uses an information set which contains only a small and quite arbitrary subset of government decisions – it is, literally, not capable of representing the system it is meant to control at an acceptable level of accuracy.

The justification for ignoring the future consequences of current spending seems to be no more sophisticated than what we pejoratively call “Treasury Brain” – the belief that a) spending departments always claim that their budget should be seen as an investment, b) that they systematically overestimate the long term benefits of current costs and c) that the best solution to a) and b) is to completely ignore any future benefits from investment or future costs caused by noninvestment. It’s just not good enough.

It is this dog’s breakfast that apparently tells us whether we are on track to meet the plotline the government has set for itself. Its three fiscal rules:

Rule 1. The current budget should be on course to be in balance or surplus by 2029/30 (‘stability rule’);

Rule 2. Net financial debt should fall as a share of the economy in 2029/30 (‘investment rule’); and

Rule 3. Some types of welfare spending must remain below a pre-specified level (the ‘welfare cap’).

And if that gruesome threesome doesn’t set your heart racing about the country’s future story, you are not alone. Because the people banging on about fiscal space will always push for governments to spend less, regardless of the circumstances. For instance, at the height of the pandemic in December 2020, Moody’s were saying this (with excess deaths already estimated at over 30,000 in the UK by Public Health England at that point):

However, compared to the government’s March budget (that was quickly overtaken by events), there are some initial signs that fiscal policy outside of investment is likely to be less expansive than previously announced. What remains unclear is whether this ambition will be able to withstand the political pressures that seem to be inevitable given the government’s previous commitments. Even before the Spending Review, longer-term spending commitments for health, education, and defence had already been announced. Together, these three areas account for around 60% of total expenditure.

One of the themes from the Funding the Future conference was, to quote from Educating Rita:

“There must be better songs to sing than this”.

I agree. One possibility is proposed by the New Economics Foundation, who suggest categorising policies by their level of multiplier effect (ie how much economic activity is increased by an increase in spending) and replacing fiscal rules with a “fiscal referee” who can make a more holistic judgement addressing many of the omissions Dan Davies complains about. This would then lead to a discussion rather than the compliant/non-compliant cliff edge we currently have, and allow the government to make its case. And it would not automatically incentivise cuts like the current system does.

Other possibilities exist. As Dan Davies describes it:

any accounting system is a mental prison of some kind, and one which needs to be periodically escaped from

In my view, a jailbreak is long overdue.

The IFoA’s AI Manifesto – a response

01. October 2026 · Write a comment · Categories: Actuarial, AI, Cartoons, Pensions, Professions, Regulation, Risk, Science Fiction
A no entry sign over an image of a reverse-centaur where the robot is in control

Last month the Institute and Faculty of Actuaries (IFoA) launched what it calls its “AI Manifesto“. I thought I would take a look.

The Manifesto is admirably short and to the point. Its pitch can be summarised by this final line of its Introduction and context:

For the benefit of society, we encourage those responsible for developing and deploying AI systems to ensure that there is an actuary in the loop.

Think of it as a massive “Gissa job” plea to those companies currently investing in AI systems. But what sort of jobs are we pleading for?

The idea is that actuaries have been working with models which use machine learning for many years, that we have statistical expertise, are comfortable working with data, used to stress-testing systems and also to looking at the risk appetite of a business as a whole.

I think the first hint that there might be a problem here is when the Manifesto moves onto ethics and the public interest. As it says:

Actuaries help teams balance commercial aims with societal expectations, offering constructive solutions rather than imposing limits.

It is difficult to square this with Anthropic‘s recent IPO filing, where they warn its technology may pose “existential risks to humanity”. I am pretty confident that Society expects not to be put at existential risk. So if Anthropic thinks its technology is that dangerous, there is no balance to be found between commercial aims and societal expectations here. Societal expectations are paramount. And limits will need to be imposed rather than “constructive solutions”.

But, in my view, the main problem here is what “real world” the IFoA feel they are serving. This comes through in the role for “professionalism” they envisage:

Projects with sound validation, honest documentation and clear accountability clear governance more quickly, avoid late-stage rework, and fail in testing rather than in production. Professionalism is what lets an organisation say “yes” to AI with confidence – and protects it from the adverse outcomes that would otherwise force it to say “no”.

This, remember, is in an environment where the latest large language models like Anthropic’s Claude Opus 5.5 or OpenAI’s GPT-6 Astra, are proprietary models where validation, documentation, accountability and governance are all just what they let you see. A framework which allows companies to say yes with misplaced confidence is unlikely to be serving the public interest. And translating a capacity to apply a regulatory model to an insurance company where you have all the data and can make informed judgements about the things that you don’t know (Actuaries are used to working with large, imperfect, transactional datasets subject to audit and regulatory scrutiny. We clarify what data is needed, what quality level is acceptable, and how data should be checked, reconciled and monitored over time) is emphatically not the same as working with a model where you don’t know precisely how it works, cannot accurately predict the errors it will make and do not have enough time to look in all of the places you need to.

Take bias for instance. The Manifesto says:

We examine how a model behaves for different groups and sub-groups, test whether training data is representative of the people the system will affect, and help manage the biases we find responsibly through fairness metrics for predictive models, and output audits and red-teaming for generative systems.

Nita Farahany in the law and policy course documented in her Thinking Freely Substack demonstrates the limits of that bias management in practice, and this is Cory Doctorow on red-teaming (from Red Team Blues):

That’s the problem with blue teaming it—you need to be perfect, while—The red team only has to find a single error

Yes, that’s right, the IFoA haven’t even understood that they are on the Blue Team when it comes to AI errors, ie the one supposed to be stopping as many of them as possible. The Red Team breaks into systems by exploiting a single vulnerability. That is comparatively easy. The Blue Team needs to reassure its clients that it has removed ALL vulnerabilities.

That will be the role of any “actuary in the loop”. And the chances are it won’t be possible. The three examples given by the IFoA are as follows:

A generative AI customer assistant
An insurer wants an LLM-based assistant to answer customer queries. An actuary designs the evaluation
framework: a curated test set of realistic queries, explicit tolerances for error, escalation rules for high-stakes topics, and live monitoring of answer quality. The project clears internal governance at the first attempt because approvers can see exactly what “good enough” means and how it is evidenced.

A machine-learning decision model
A lender deploys an ML model to score applications. An actuary tests behaviour across customer sub-groups,
identifies a data artefact that disadvantages one segment, and works with the data science team to correct it. The documentation of that judgement becomes the centrepiece of a constructive conversation with the regulator.

AI risk across an organisation
A Chief Risk Officer faces AI adoption in a dozen business units at once. An actuary helps articulate an enterprise AI risk appetite, maps where the organisation depends on the same foundation models, and builds board reporting that distinguishes experimentation from deployment. The organisation accelerates adoption in low-risk areas precisely because it now knows where the high-risk areas are.

None of these examples look unreasonable in isolation, but, to justify the eye-watering expense of this technology, it will need to increase speed and capacity significantly. So, rather than a single generative AI customer assistant, imagine a client relationship management role where a junior actuary or senior student manages 10 times the current expected number of clients at the same time, because the generative AI customer assistant makes that possible?

Will the skills of an actuary versed in the intricacies of Solvency 2 or pensions regulation be so revered by your client that you would be given more than enough time to rubber stamp a system serving 30 or 40 clients? How long do you think you’d need to check the output on all of them properly? I think it unlikely you will be given that long.

This is the risk of becoming a reverse centaur, where you work for the assistant rather than the other way around. And my view is that the actuarial education system is positively encouraging the development of reverse centaurs amongst our current students.

It gets worse. The early evidence suggests that working with AI systems encourages rubber stamping. In one study of AI-generated audits with an auditor in the loop, 150 of the 160 reports were submitted without any changes made by the human. The performance of humans in the loop may also be adversely affected by working with AI systems, leading to “agency decay”:

Agency decay is the gradual erosion of a person’s ability and willingness to observe carefully, think independently, choose deliberately, and act responsibly. It does not arise because AI is inherently harmful. It arises when convenience becomes the default setting for cognition. A tool that first helps us think can begin to think around us, then for us, then without us noticing what has been weakened.

There are lessons to be learned from other professions who have been working for much longer with expert systems. The aviation industry for example. As Craig Bright, Co-Chief Operating Officer at Barclays has posted:

So the test shouldn’t be whether a human appears somewhere on the process map.

It’s whether the system remains understandable, controllable and recoverable when the agent is wrong, uncertain or unavailable.

Safety doesn’t come from leaving a human in the loop.

It comes from designing a system the human can still control.

The IFoA’s position appears contradictory. On the one hand, they clearly don’t believe that AI is as powerful or as dangerous as Anthropic are claiming in order to push up the IPO share price, or I assume they would not be pushing so hard to get an actuary on every team “developing and deploying AI systems”. I think they are right not to believe it, for a number of reasons, but Naomi Alderman’s AI predictions: seven ways you can tell useful thinking from sci-fi fantasies gives probably the most entertaining ones.

On the other hand, if this is just another technology for which the price is going to need to adjust to reality at some point, we need to be very careful as a profession not to be part of the mob urging their clients to adopt it at a faster rate than they would otherwise. Particularly if it is damaging our own professionals in the process.

AI in some form will survive the inevitable crash, but the shape and scope of the wreckage that will be left behind is currently hard to predict. Let’s be careful what we wish for.

Three names for the same idea

15. September 2026 · Write a comment · Categories: Actuarial, AI, Cartoons, Education, Professions
Source: https://xkcd.com/2347 This work is licensed under a Creative Commons Attribution-NonCommercial 2.5 License

Last week I suggested that the period of disruption we have entered would be likely to impact the relationship between society and its institutions and professions, and likened those institutions and professions to the models behind the models we normally focus on. I suggested that this has serious implications for actuarial education, amongst much else.

This week I want to expand a little on what I meant by that.

First, the impact of AI. My view of AI is that I don’t believe most of the claims coming out of OpenAI and Anthropic when noone else gets to see the data and they clearly have a vested interest in playing up the capabilities of their systems. And I certainly don’t believe the fantasy economics the Magnificent 7 are currently indulging in is sustainable (see Ed Zitron’s Hater’s Guide to Circular Financing for more on this). But I do believe that there is a lot of political power with an increasingly out of control stake in it continuing, which makes the quote often attributed to Keynes (“Markets can remain irrational longer than you can remain solvent”, although it may have been A. Garry Shilling in 1999) even more relevant here.

However, what cannot be denied is that the existence of AI is accelerating conversations which we should have been having a long time ago. Last month, MIT issued a report called AI and Education, which turns out to be more a philosophical musing about what we should be trying to achieve in universities, and what expectations we should have of both staff and students. It’s not a long report, but for the time poor amongst you there are two passages from this that I’d like to highlight here.

First, on the expectations we should have of students:

Leaning into learning means creating a new “social contract” between teachers and students. All of us who teach at MIT will need to be prepared to help students understand both that the process of education is necessarily a productive struggle, and that the most important product of their education is not a GPA or a diploma but themselves: their personal growth and intellectual maturity and the development of their own imagination, insight, and judgment.

Instilling these attitudes needs to become a central task for every educator, so that our students know not only what they should learn, but also how they should learn and why learning matters. We need to help them develop metacognitive abilities to think about thinking, to engage in reflective practices, and to enhance their sense of personal agency. This will require both dedication and fresh preparation on the part of instructors.

Remember that social contract idea, because it will be coming up again.

Second, on what kinds of assessments we need to move to in anything we still want to be able to refer to as education:

Rather than simply “AI-proof” current methods of assessment, instructors need to revisit what they really want students to know and devise assessments that foster, or even include, the kind of productive struggle that builds durable understanding and mastery.

We urge instructors to consider forms of assessment that are less vulnerable to AI, and more valuable for learning, such as oral exams, semester portfolios, and out-of-class assignments paired with in-class conversations. This likely means resources such as TAs and class time will become more central to evaluation.

and this:

To match the shift towards assessment methods that aren’t vulnerable to AI, instructors need to increase the role of experiential and project-based learning. To encourage this kind of creative teaching, MIT should support the development of teaching skills and practices for all instructors, and recognize contributions in this domain.

As someone who has already spent some time advocating just such a move in higher education, I feel quite vindicated. But back to that social contract point. I am in a WhatsApp group called Actuarial Change in Education (A C E), where we have been discussing this point for a while, but couldn’t agree what to call it. Culture was one proposal, but I think we all agreed on the MIT point that it should be aiming to ensure:

our students know not only what they should learn, but also how they should learn and why learning matters

And then, like buses, a third take on this point came from, of all people, Seth Godin in Two schools of thought about school:

One kind of school is about culture, compliance and community. We indoctrinate kids, for better or worse, in what it means to be in this world we’ve built, and how to do it together. We teach them how to be kids, how to be citizens and hopefully, how to be adults.

And the other kind of school is about the content. All the ‘R’s. The software that goes with the operating system we taught in the other school.

The first paragraph is what MIT and my WhatsApp group are referring to. It is the bit which has been assumed we shared in common for too long, but is the most important bit. A shared culture or social contract would allow us to meet and discuss things within a shared set of assumptions about how we can get agreement and take action. It is what has been holding back COPs and peace talks and politics more generally for my whole lifetime. It is also how we decide what we put in the second paragraph, which is necessarily stuff which will be being tweaked constantly and be much more short-term in focus..

The second paragraph is what the Institute and Faculty of Actuaries respond with when you ask them about education (I am only picking on them as this is the one I am most familiar with). Curriculum. Syllabus items. Suggested number of hours study. Past papers to practise for the traditional written exam.

Seth Godin calls this software. It is content. It is largely ephemeral, as the needs of the model we operate in and the tools we are given to do so change all the time. Little of it will survive a decade let alone a career. It does not encourage looking up and wondering if there is a better way. It is full of short-term deadlines and urgency and immediacy. Which tends to get confused with importance.

But the software will never surprise you or deal with situations noone has ever met before. If it is not in the underlying model, it will not turn up in the software. The important education is in the operating system. That is what needs to be reformed.

As I said, there is a WhatsApp group discussing such questions regularly. We are a small group currently, but would welcome new members interested in these questions. If you are, then respond to me in the comments section below or via a LinkedIn message if you are reading it there.

A little more information about A C E is set out below:

What is A C E?

Actuarial Change in Education. A place to discuss where change is needed in how actuarial students learn and the changing landscape they are preparing themselves for. A place to argue about how we teach, and how we get students to care about why they are in education and question the assumptions underlying it. A place to share thoughts and ideas you have had or seen about education, actuarial or otherwise. A place to collaborate with others equally concerned about the future of actuarial education. A place to plan and deliver communications with the wider actuarial community.

What are A C E’s aims?

To have an impact on actuarial education by providing a place for developing actionable ideas about education more generally and how they can be applied to actuarial studies in particular. The hope is that these will then be a useful resource for when the need for alternatives gains a wider acceptance.

Who can join A C E?

Anyone interested in the development of actuarial students.

Models of the first and second kind

09. September 2026 · Write a comment · Categories: Actuarial, Economics, Education, Health, Professions, Science Fiction
Copper parts of a vehicle have been mounted on a much larger wooden structure in an exhibit at the Vatican Museums
My photograph of an exhibit at the Vatican Museums

Worldbuilding in science fiction has certain rules, or so I thought. And the first of these according to some sources at least seems to be consistency. According to the Quirkworthy website:

The basic message is very simple: your world must be internally consistent. Doesn’t matter if it’s science fiction or fantasy, and it makes no odds if it’s a novel, game, or opera. At all times, your world must be consistent to itself.

So when Jeff Noon and Steve Beard came to the Birmingham Science Fiction Group a couple of years ago, I pricked up my ears when one of them (Jeff I think) said that the first book of the Ludwich Chronicles, Gogmagog, which they were discussing, was not consistent. They were following Cady Meade’s viewpoint which journeyed 60 miles along the river Nysis inside a ghost and the focus was to set a particular mood. It was built from the bottom up.

This idea appealed to me then, and has, in my view, become a more and more important idea recently. We all build our views of the world from the bottom up based on our own experiences and they are frequently inconsistent with those of other people. The problem comes when we stop accepting that and treat our models as, in some way, top down and universal.

I had a great example of this a few years ago in the Vatican Museums. Some bits of bronze they reckoned to be parts of a cart buried in an Etruscan tomb had been discovered and were then displayed on an entirely speculative wooden model of a cart (to be fair to them, the Monteleone chariot dated to around 530 BC, was a rather more complete find and the wooden model looks like it was modelled on that). The trouble is, once you have seen the pieces on the cart, it is difficult to think of them in any other way. Rather like some of those early dinosaur reconstructions and possibly unhelpful in a field as fast-moving as archaeology.

Source: The book of the great sea-dragons, Ichthyosauri and Plesiosauri, [gedolim taninim] gedolim taninim, of Moses. Extinct monsters of the ancient earth. With thirty plates, copied from skeletons in the author’s collection of fossil organic remains, (deposited in the British museum.) by Hawkins, Thomas, 1810-1889
https://archive.org/details/bookgreatseadra00hawk/page/n5/mode/2up

We do this everywhere. Every competitive market is a model, as relentless in its pursuit of victory within the rules (all of which must be stress-tested as to how they are going to be applied by each individual referee in each individual encounter) as the Premier League. Sometimes if you question some of the underlying assumptions behind some of these models, you are told that is just the way the world works.

But the way the world works is opaque. Novelists have attempted to show us glimpses of it for centuries. Historians are sometimes able to agree about how it has worked in a specific geography for a short period of time. Occasionally, as a result of our own experiences, we may feel that we have managed to grasp some truth about the world and how it works in our own lives. But often the clarity slips through our fingers when we throw more light on it. The clarity certainly never seems to last that long. For clarity you often need an organisation signed up to a model.

Much has been written in recent years about our dependence on models and our inability to look outside them in many cases, most notably Escape from Model Land by Erica Thompson. This has mainly focused on the models we make to try and understand the world. But there also models behind those models which effectively determine what kinds of models of the first kind we create. These are the organisations and institutions we look at the world from.

It was in a time of rapid transition and uncertainty much as our own is that the groups of organisations we call professions emerged in the nineteenth century. The model of the second kind they adopted was partly that set out by Daniel and Richard Susskind in their book The Future of the Professions.

According to the Susskinds, there is a Grand Bargain between society and the professions, which means (and I am paraphrasing a little here), in return for professions providing:

  • Expertise, experience and judgement;
  • Delivered affordably, accessibly, reassuringly and reliably;
  • With knowledge and methods maintained and kept up to date, members trained, standards and quality of work enforced and only appropriately qualified individuals allowed in;
  • Acting honestly and in good faith; and
  • Putting clients’ interests ahead of their own.

In return, society will give the professions:

  • Respect and status;
  • Exclusive rights to perform/provide socially significant activities or services; and;
  • Independence to decide how they do it and how much they can be paid for it.

What professions expect from society is jealously guarded. My own profession’s uneasy relationship with the universities can plausibly be traced back to Newton’s influence on what mathematics could be taught there, therefore denying actuaries the status they craved without setting up their own professional bodies to supply it. The last attempt to allow an allied profession to carry out limited medical activities ended with the Leng Review effectively strangling it, egged on noisily by the BMA. And it is not hard to find articles defending increases to executive remuneration way beyond that of the people who work for them as the norm.

What society expects from professions often gets less attention directed at it. Four of the five bullets above tend to form part of the professional code of most professions in some form, although there are frequent fines of large professional firms for failures in the standard of the services provided.

But what of delivering professional services affordably, accessibly, reassuringly and reliably? What has tended to happen, as the group of people requiring professional services has widened, is that the affordability and accessibility have been increasingly called in to question. And online self-service looks likely to be attempted for a wider and wider range of services as time goes on.

So is the professional structure we have going to remain? If the experience of the nineteenth century, when most of these professions emerged, is anything to go by, the level of disruption will vary hugely by profession and institution and the period of transition could be very long in some cases. But it seems clear that such a period has begun. This has serious implications for actuarial education, amongst much else.

And whatever replaces them is likely to be built from the bottom up.

Normals: new and otherwise

25. August 2026 · Write a comment · Categories: Actuarial, Climate Change, COVID-19, Economics, Health, Media & Politics, Risk

Climate campaigners get very irritated with media talking about the “new normal”, pointing out, quite reasonably, that for it to be the new normal would require a degree of climate stability that we have not got, nor have any prospect of getting while we continue to argue about whether net zero (the policy, remember, which is about moving us to a position where we don’t increase the amount of carbon in the atmosphere year after year) is affordable. Poverty and health campaigners got very irritated by the call to “get back to normal” after the pandemic, when it had so clearly demonstrated how our “normal” way of organising society had failed to protect so many of its citizens. As even I could see in March 2020:

My view is that some things that must be different post COVID are already clear. I think as a society we are going to demand more resilience, for example:

  • Resilience of our health service – this means much higher levels of spending, building deliberate over-capacity into the system in normal times;
  • Resilience of our food supplies, for example strengthening domestic supply chains;
  • Resilience of our population, so that we do not have 1.6 million food parcels needing to be given out in a year by the Trussell Trust, in the absence of a pandemic, for instance; and
  • Resilience of our infrastructure – to everything from floods to banking crises to pandemics to storms and heatwaves.

A picture circulating at the time of a Santiago apartment block with a projection onto it of “No volveremos a la normalidad porque la normalidad era el problema” (we won’t get back to normal because normal was the problem) was actually taken the previous year when the Chilean President had used the word “normality” to justifying lifting the state of emergency, leading to large protests:

Source: https://x.com/inesmorsantos/status/1190192093165211648?ref_src=twsrc%5Etfw%7Ctwcamp%5Etweetembed%7Ctwterm%5E1190192093165211648%7Ctwgr%5E%7Ctwcon%5Es1_&ref_url=https%3A%2F%2Fodi.org%2Fen%2Finsights%2Fcovid-19-we-wont-get-back-to-normal-because-normal-was-the-problem%2F “We won’t go back to normal, because ‘the normal’ was the problem.”

But we all have trouble with fast moving situations, or even just moving situations. Remember the media coverage in the early weeks of Covid utterly focused on the number of cases going up each week rather than on the underlying pattern this showed, which had been predicted and could be responded to.

In the Institute and Faculty of Actuaries syllabus, subject CS2 Risk Modelling and Survival Analysis has 20% of its content on dealing with our trouble with moving situations via time series:

It boils down to a number of tricks to turn our moving situation into a non-moving or stationary situation, which we can then analyse.

So not being able to deal with things that aren’t stationary is entirely normal. Indeed the Standard Normal Distribution, originally developed in response to a gambler’s query and another thing that we spend a lot of time transforming other distributions into if we can, has a fixed mean of 0 and a fixed variance (a measure of how far things stray from the mean) of 1:

Outside statistical analysis, we use different techniques to make things appear normal. Denial, ie ignoring the departure from normality, is a big one. When this one is no longer tenable in the face of mounting evidence that we are not in Kansas anymore, we often try and depict where we have got to as normal now. This will have been a good strategy in the past: once a natural disaster has passed, taking stock of what has been lost and what remains and working within these new parameters makes more sense than hankering back to how things were before the disaster struck.

However this strategy comes unstuck when one or both of two things are true:

  1. The disaster, or change-event more generally, is not a one-off but a continuing process.
  2. The way we were living before the disaster made the disaster more likely.

So, on climate change, both 1 and 2 are true. On the pandemic, 2 (if you define the disaster as the number of people who died during it, which I do) is clearly also true. And, although it may not be a continuing process, none other than the former Chief Scientific Advisor to the UK Government regards another pandemic as “inevitable”.

We have built systems to look at past disasters and, in particular, try and guard against 2. The Covid Inquiry is working its way through the evidence about the pandemic, and has already produced definitive evidence on how we could do things differently for those who wish to read it, but the reporting of its work is patchy, with sometimes unhelpful fixations on the wrong thing.

And denial tends to fight back as our go-to defence mechanism, which is why we often seem to need to have Public Inquiries about the same things again and again and again.

It is all quite normal. That is the problem.

The inequality of redundancy

05. August 2026 · Write a comment · Categories: Actuarial, Birmingham, COVID-19, Economics, Health, Pensions, Science Fiction

In October 2009, while working as a pensions actuary in Birmingham, I attended the Institute and Faculty of Actuaries’ Joining Forces in Mortality and Longevity multidisciplinary conference at the Royal College of Physicians in Edinburgh. The canapés were excellent and, of the plenary speakers, I particularly remember Rudi Westendorp from Leiden University who, in response to the question of whether it was nature or nurture that was responsible for longevity, said it was both.

It was also the first and last time I have used a sleeper train. I had a trustee meeting on the Friday morning south of London and decided that it would be a good idea to get the sleeper from Edinburgh to London on the Thursday night. I hadn’t really researched it in advance and I suppose I was expecting a cabin to myself, but it turned out I was in fact sharing with a German guy who looked like he rode the overnight trains of Europe all the time. There was a set of bunk beds which took up most of the cabin and my memories of that night were of my being thrown around every time the train took a bend down the East Coast mainline, not sleeping a wink while my room-mate snored soundly throughout and then of his taking an annoying amount of time to brush his teeth in the morning. The trustees did not get me at my best.

But I digress. The session from the conference which has stayed with me ever since was by Eugene Milne from Newcastle University. As the blurb for his session A new model of mortality and survival stated:

A new mathematical model mimics patterns of mortality in species from nematodes to humans. Net risk, it suggests, arises from two components, an interactive element accounting for historical falls in human mortality, and redundant decay which shapes lifelong risk. This suggests “intrinsic ageing” is unrelated to the slope of the mortality curve, and is unlikely to have altered for humans in recent centuries.

Eugene Milne went on to become Director of Public Health for Newcastle and took them through the pandemic. The full paper in the British Actuarial Journal from 2009 is here.

At a time where actuarial mortality modelling consisted either of projecting observed trends into the future (with no real explanatory power for observed features like cohort effects) or attempted to project a range of causes of mortality based on the likely progress in treating them (which is limited by the quality of information on death certificates), Milne’s model stood out for starting from a biological model of gradually diminishing redundancy within the body. He recognised its limitations, as he said at the time:

As a bridge between actuarial and biological approaches to ageing it falls, as yet, between two stools. The examples in this paper are generated by probabilistic computer modelling. This is clearly not adequate for
actuarial use and the model needs to be developed further to provide a form that will serve that purpose. On the other hand, it describes (as noted in section 4) a biology that would be ‘needed’ in order to construct organismal risk as it is observed. This it does well, providing a coherent account of why mortality patterns appear as they do. Yet its theoretical ‘biology’ is at odds with currently favoured theories of ageing. If the biological quantity described in the NBM [nested binomial model] as redundancy exists, we do not yet know what it is, nor why it should appear to act in so sequestered and consistent a fashion.

But the idea stayed with me, and perhaps that made me particularly receptive to Richard Murphy’s recent article about allostatis, which he describes as “the way living organisms survive by continually adapting to the changing world around them”. As Murphy points out:

If something breaks, a wealthy person can simply have it repaired or replaced. If travel arrangements fail, alternatives can be found. If income is interrupted, there are savings to draw upon. Problems remain, because they always will, but they rarely threaten the stability of everyday life for a wealthy person. In other words, the amount of adaptation required of them is reduced.

There have been a number of attempts over the years to draw attention to how your level of income affects your ability to adapt to your environment.

In George Orwell’s Down and Out in Paris and London, he describes the effect of poverty:

For, when you are approaching poverty, you make one discovery which outweighs some of the others. You discover boredom and mean complications and the beginnings of hunger, but you also discover the great redeeming feature of poverty: the fact that it annihilates the future.

However Orwell has been criticised, as the temporary tramp that he was, for deciding that meant the end of anxiety. More recently, Jack Monroe used Terry Pratchett’s example of Captain Vimes’ boots, to illustrate how what life costs depended on how much you already had:

A really good pair of leather boots, the sort that would last years and years, cost fifty dollars. This was beyond his pocket and the most he could hope for was an affordable pair of boots costing ten dollars, which might with luck last a year or so before he would need to resort to makeshift cardboard insoles so as to prolong the moment of shelling out another ten dollars.

Therefore over a period of ten years, he might have paid out a hundred dollars on boots, twice as much as the man who could afford fifty dollars up front ten years before. And he would still have wet feet.

Jack developed the Vimes Boots Index to highlight inflation in basic food products, which then persuaded the ONS to launch a shopping prices comparison tool.

But to return to the idea of redundancy: it is clear that if you have, say, redundant money which can be brought in to play when things go wrong, then things generally go much better for you. But imagine if you have used up any redundancy quite some time ago and therefore any reverse is a potential crisis. Your stress levels will be much higher all of the time, as your body tenses for the battles it knows lie ahead about anything from a car that won’t start to a boiler that packs up to a hike in the price of bus fares or the basic food products on the Vimes Boots Index. Day in, day out.

At the moment one of the differentials in redundancy is around heat stress. Housing which cannot be cooled without electricity which cannot be afforded. Day in, day out, for much of the summer so far. But in winter the same housing will need far more heating because its occupants cannot afford capital investment like insulation.

And the reason for this enormous inequality in the level of redundancy, is of course wealth and income inequality. The Equality Trust sets out the statistics as follows:

The UK has very high inequality of income compared to other developed countries; the 9th most unequal incomes of 38 OECD countries (OECD, 2022).

The UK’s wealth inequality is much more severe than income inequality, with the top fifth taking 36% of the country’s income and 63% of the country’s wealth, while the bottom fifth have only 8% of the income and only 0.5% of the wealth according to the Office for National Statistics.

And this stress, day in, day out, means that people also die earlier, as shown below in the House of Commons Library research briefing on Inequalities in life expectancy, using ONS figures (and honestly they’re not picking on Blackpool):

And yet the most likely way for you to hear the word redundancy is when some company is planning to get rid of some of its workforce.

The trouble with people is that you don’t need them until you do, as Jonn Elledge has written here:

There are substantial differences between the US military and the NHS: but what both have in common is that, when they need something, they really do need it. The entire notion of “efficiency” is misplaced.

Similarly, in Nate Hagens’ latest interview, with biologist and biophysicist Olivier Hamant, they explore:

…how cheap, abundant energy allowed human societies to substitute the “safety net” of resource abundance for the safety net that living systems actually rely on: diversity, redundancy, and cooperation.

Until the resources stopped being so abundant.

And then there is the kind of workplace sustained redundancy programmes create, as Seth Godin says in a recent blog post:

“How few people can we get away with?” That’s a question many bosses think hard about. Automate. Streamline. Improve productivity and take humans out of as many tasks as possible. It’s a time-tested way to create profits and to increase a certain kind of reliability. Claude Code is popular with many organizations for precisely this reason.

“How can we include more people into this process?” is a less popular but often more valid way to create value. In an organization that is in the business of bringing insights, humanity and care to problems, more involvement from people creates better outcomes. The challenges make it far more valuable.

Which sort of organization would you like to work for?

And what sort of society do we want to have? One that supports the people in the bottom quintile with no redundant capital, so that they might have the resources available to maintain the stability of everyday life and reduce the adaptation required of them? Or one which keeps demanding more and more from those it allocates less and less of national wealth to. To quote from another Orwell book, Nineteen Eighty-Four, from the mouth of O’Brien, Big Brother’s Grand Inquisitor:

If you want a picture of the future, imagine a boot stamping on a human face—for ever.

I think we can all come up with better pictures of the future than that one.

Nutted by Newton

03. July 2026 · Write a comment · Categories: Actuarial, AI, Economics, Education, Professions, Risk

Isaac Newton may or may not have been nutted by an apple. His friend William Stukeley, whose memoir of Newton was the source of the story, states it as follows (spelling and punctuation from Stukeley’s manuscript at the Royal Society):

“why should that apple always descend perpendicularly to the ground,” thought he to him self: occasion’d by the fall of an apple, as he sat in a comtemplative mood: “why should it not go sideways, or upwards? but constantly to the earths centre? assuredly, the reason is, that the earth draws it. there must be a drawing power in matter. & the sum of the drawing power in the matter of the earth must be in the earths center, not in any side of the earth. therefore dos this apple fall perpendicularly, or toward the center. if matter thus draws matter; it must be in proportion of its quantity. therefore the apple draws the earth, as well as the earth draws the apple.”

However there may be stronger grounds for believing that Newton nutted English mathematics and, as a result, the actuarial profession.

I thought all this to myself, occasion’d by participation in the Actuarial Teachers and Researchers Conference (ATRC) this week. It was great. There were lots of interesting talks, and people were very engaged and respectful in how they discussed them. There was a lot of expertise in the room about both education and actuarial practice and theory. It felt like it was a room that could take on difficult topics and make progress in tackling them.

There were several calls for the need to change assessment, one of them by me, whether due to large student cohorts and the difficulties of engaging them, or the ease with which AI can duplicate examination solutions or because the credentials provided by those assessments seem increasingly irrelevant to the actual actionable skills, knowledge and experience required to operate successfully as an actuary. And there was an equally robust response from the Institute and Faculty of Actuaries (IFoA) to the effect that assessments would not be changing significantly in the latest changes due to come in by 2029.

My favourite presentation was by Angus Macdonald, about his recent short paper on Newton, Leibniz and Actuarial Science. It conjectures that the argument between Leibniz and Newton over who invented calculus resulted in the stagnation of mathematics in English universities and meant that actuaries needing to gain respectability for their advice were forced to create a professional body (in company with architects, accountants and engineers) rather than rely on universities to develop actuarial thinking as happened in continental Europe. This stagnation (Macdonald quoted G.H.Hardy, talking about the Cambridge mathematics exams, saying that they had “. . . effectively ruined serious mathematics in England for a hundred years”) was felt across the whole Anglosphere, with attempts to create institutions with equal status to the traditional universities leading to the establishment of business schools in the USA in the late 19th and early 20th centuries.

This explained a lot for me. The suspicion of our actuarial courses amongst some of our mathematics colleagues at times at Leicester. The occasionally uneasy relationship between the IFoA and the accredited universities where students can gain exemptions from their exams. The horror from some at the very idea that an actuary might qualify as a result of university courses, like they can in the Netherlands, for instance. ATRC seemed the appropriate place to explore these ideas.

Angus Macdonald considers that the Newtonian and Leibnizian branches have nearly rejoined, but I feel that our actuarial education system still suffers from the long shadow cast by these two 17th century gentlemen and their personal enmity. The ferocity with which some within the profession opposed what they saw as a threat to the popularity of the fellowship qualification posed by the new chartered actuary designation was hard to explain when the number of practising certificates issued for chief actuaries, pension scheme actuaries, etc was only around 1,100 (out of over 17,000 fellows), but if the whole profession originated from a highly developed form of status anxiety compared to academics, it becomes much easier to understand. As they say, the apple doesn’t fall far from the tree.

Now, nearly 180 years on from this “unstoppable” “rush to respectabilize” (according to Jeremy Paxman) I think it is legitimate to ask:

  1. Why does the IFoA still insist on detailed accreditation of individual university courses on basic mathematics, statistics and business economics? Do they not trust them to teach them right? Do they think that there is something better about learning these universal subjects via a bespoke course created by a tiny professional body with 34,000 members globally?
  2. Why are they so concerned about what assessment methods the universities use? Do they consider that they have more educational expertise on this than some of the leading universities in the world?

I have written previously about the changes in assessment I think are necessary and floated some ideas about where actuarial education might sit to accommodate these. My view was that the IFoA, faced with the need to innovate at all levels of its education system at a time of great uncertainty, might wish to get out of the foundational mathematics and business education tuition of the core principles subjects and leave this to the university system, in a truly Leibnizian way.

However there is a problem. The university system is facing a tough time. UK higher education is shrinking according to the UCU branch at Queen Mary’s, which has set up a a live page of all the redundancies, restructures, reorganisations, and closures taking place across the UK Higher Education (UKHE) sector. The numbers from the Higher Education Statistics Agency (HESA) bear this out, with overall student numbers falling, driven by a 10% fall in overseas entrants with a non-European Union permanent address.

In actuarial science, the list of UK universities offering accredited courses has not changed for some time, and it is a small group within the university system. These courses must be seen as vulnerable. I don’t think it is a coincidence that it is so hard to persuade one of this tiny band to run the ATRC. This week’s was the first since 2021 and the list of past hosts is not very long:

Source: https://atrc.le.ac.uk/

The IFoA needs to ask itself what it would do if there were some high profile closures of actuarial courses and whether perhaps it might be time to protect the core mathematical education arrangements for actuaries through a broad brush accreditation of a wider range of courses, rather than the very detailed accreditation process it currently carries out on core principles subjects.

I also doubt that it has the capacity on its own to significantly restructure its assessments beyond the objective based assessments it has developed to work within its proctored environment.

The core practice and specialist subjects leading to fellowship are the areas where actuaries have, rightly, the strongest views about content and assessment structures. It surely makes sense that the expertise within the profession and the university system are pooled in a joint endeavour rather than risking both failing independently.

In contrast to Newton, who had quite a temper, Leibniz was famously parodied by Voltaire in his Dr Pangloss character in Candide who believed that “All is for the best in this best of all possible worlds.”

Let’s not be too Panglossian about the current state of actuarial education.

How Not To Be A Reverse-Centaur – Part 2

16. June 2026 · Write a comment · Categories: Actuarial, AI, Cartoons, Economics, Education, Professions, Risk, Science Fiction
A no entry sign over an image of a reverse-centaur where the robot is in control

As I discussed in How Not To Be A Reverse-Centaur (Cory Doctorow defines a reverse-centaur as “a machine that is assisted by a human being, who is expected to work at the machine’s pace”) our actuarial education system needs to change. This has been true for some time, but the development of AI systems has both demonstrated why more clearly and accelerated the timeframes over which action is needed. As I said in December last year:

A large part of the education of the future will need to be about equipping us all to understand what we now have access to and when and how to access it. We will all have different things we are interested in, or end up involved with and needing to be educated about. It will be up to each of us to decide which things are worth the difficulty of learning, but to make those decisions we will need education that can support the development of judgement.

For education institutions, the question will be what is not worth the difficulty of learning? Credentialising based on now relatively meaningless assessment methods will not cut it. This is where the confrontation with employers and politicians is likely to come. Essential skills and their related knowledge will be better developed and assessed via more open-ended project work and online assessment of it to check understanding. These will need to become the norm, with written examinations becoming less and less prevalent. Not because of fear of cheating and plagiarism, but because an outcome which can be replicated that easily by AI is not worth assessing in the first place.

Suppose you were to ask someone who had never seen our academic system that, in order to assess whether someone else we wanted them to employ in their organisation would do a good job for them, we would:

  1. Award marks for what has been written in answer to academic questions about what the students can remember unaided about the content of their lecture courses and reading lists with a biro on a pad of paper perched precariously on a tiny wooden table surrounded by hundreds of other similar scribblers, for a set period of time as minders wandered the floors like Victorian factory owners.
  2. These marks would be based on marking criteria they would never see.
  3. We would then get together in a secret huddle for a couple of months to mark all these scripts, check them and “moderate” them to check they were in line with the previous years’ exercises.
  4. Then out would come the answer (like the Answer of 42 to the Ultimate Question of Life, the Universe and Everything in Douglas Adams’ The Hitchhikers Guide To The Galaxy).

Source: https://hitchhikers.fandom.com/wiki/42

Imagine another world where the kinds of things employers routinely do when trying to decide whether to take on someone as a new member of their organisation are routinely part of their assessments at university, eg:

  • Watch them operating in groups working on a task.
  • Ask them to describe how they would go about tackling a particular problem.
  • Get them to show they understand the implications of a piece of work they have carried out and can explain what assumptions it is based on and can justify them.

With assessments of performance which are not opaque academic exercises but descriptions of performance made as the tasks are being carried out evidenced by video if necessary. In other words, transparent enough to whoever might want to work with these students next to eliminate the gap between their credentials and and their actual actionable skills, knowledge and experience.

Assessing in this way will be highly demanding, for both the students and the assessors, but it means the final assessment is done with the student present and, with careful probing from the assessors, who will obviously need to have done a close reading of the project work beforehand, confidence will be high about the abilities demonstrated by the process. And it removes the need for all the time currently spent on the Victorian factory owner process – imagine what could be done if the whole of May, June and July wasn’t spent marking scripts!

A move like this will significantly change what is taught. There will be no point presenting lots of information in ways which make retention easier for students (the so-called bookwork” questions currently in most exams). This makes sense when you think about how much of the content of your courses is accessible to you now, even on a vocational course like actuarial science, as opposed to skills developed. Instead the focus will be more on conversations with students for the purpose of developing their understanding of a meaty topic and the problems to be solved within it. Explorations of these will form the projects students will be spending most of their time working on. Checking understanding and making sure students are working within frameworks they understand and can talk about will be the main teaching activity, with opportunities to practise these throughout the year. In other words, showing the ability to apply what they have understood and demonstrate the ability to reflect, make judgements and show higher levels of understanding as a result. An ability they can then apply to new problems and situations.

And notice something else about how these alternative credentials are generated? They are all social activities: from operating in groups, to presenting a piece of work to a group of assessors. Even the project work is developed through group sessions alongside individual work. If you want to develop people who can spot the gaps between models and reality you need to do it in groups, where people can test their perspectives against each other.

Compare this to the current solitary exam preparation, peppered with “revision” lectures, exam sitting and exam marking processes which comprise the predominant activities in universities between March and July each year. We are currently being funneled into an ever more solitary professional practice around LLMs. It has become quite accepted for people to do all of their thinking on a subject closeted with perhaps one, two or three LLMs but no other people and then share their analysis with an expectation that other people should read the output. Developing the skills to do these analyses will obviously be necessary, but they will never be the most important skills people learn.

Treating each other like walking databases is never going to make you new friends or influence people. However the skills you develop via an alternative more social education will.

There are examples of this approach already happening, with employers involved in the design of assessments in some cases and universities using the flexibility they have to restructure and redesign courses. But the dominant assessment system of formal exams is keeping much of this activity at the margins at the moment.

If it were to become the dominant assessment, it would also mean that larger areas of the curriculum could be examined, as the focus would not be on retention of large bodies of content but the ability to use the content to solve problems. So, perhaps, after a first year levelling up students’ ability and experience of mathematics and statistics, year 2 would have an actuarial statistics module (currently CS1/CS2) and an actuarial mathematics module (currently CM1/CM2), both dissertation based and vivaed. Then in the third year they would tackle an economics module and a business modules (currently CB1 and CB3). Perhaps in the final year of a four year MAct they would tackle a modelling and communication module (CP2/CP3) and either CP1 actuarial practice module would be offered or a professional skills module.

This would be more of a driving test style of assessment, with students working with supervisors on the comments from assessors on the original dissertations and presentations until they were up to the required standard. Students unable to make the standard after an additional year would be offered a move to a non-accredited course. The current funding model for higher education will need to change: loading students with £10,000 more debt every time they add to the nation’s skills bank with another year wrestling with these difficult skills to master was never a clever strategy, but this structure would make it even clearer how self defeating it is. If economists are accepting that the labour market no longer allocates resources effectively and are considering either a universal basic income or universal basic services, then logically university funding should follow suit.

The implications for the actuarial profession will be even more challenging. I question whether the Institute and Faculty of Actuaries will continue to consider it worth the effort and expense of reconfiguring their entire approach to exam setting, syllabus maintenance, marking, online proctoring, etc for subjects routinely taught throughout the university system for a a variety of purposes, ie the current core principles subjects of mathematics, statistics and business studies.

Whether they retain their own capacity to assess the core practice subjects is a more open question (CP1 Actuarial Practice, I imagine, will be seen as sufficiently specialist to keep in house, with perhaps a few universities, as now, accredited to run courses which can earn this qualification, however CP2 Modelling and CP3 Communications are not nearly as specialised now as they seemed when they were first introduced). I would also move the new core economics module recommended by the IFoA Economics Review Group I chaired in 2022 to replace the current CB2 module into the core practice section, as it would be more of a critical thinking module about the economic ideas necessary to underpin good actuarial work than the technical how-to subjects in core principles.

The specialist stage of the SP and SA subjects and the whole continuing education and additional certification options I would confidently expect the profession to retain, review and adapt as needed to reflect new challenges in members’ working lives. However, if CP1 is all that is needed post graduation to get to Chartered Actuary and CP1, two SP subjects and a SA subject gets you to Fellow, then suddenly time to qualification becomes much more predictable for students and their employers alike.

The risks to employers of taking on students would therefore be reduced, but also the rewards to taking them on will be much more transparent. The students we will be developing will have collapsed the gap between their credentials and their actual actionable skills, knowledge and experience. They will have:

  • Great team working skills;
  • Very highly developed presentation skills, both in writing and speech;
  • Strong IT skills and comfort working with data; and
  • Clarity about why they are in an organisation and a drive to use their skills to solve problems.

They will also have developed the four capabilities set out in Carlo Iacono’s Teach Judgement, Not Prompts:

  1. Epistemic rigour. They will be more likely to spot when a system or model is over-confident given the evidence, after an intense experience of interrogating models and evaluating evidence in their courses.
  2. Synthesis. They will be able to integrate different perspectives into an overall understanding, through dissertation development, defending a position, understanding weaknesses in a position and adjusting accordingly.
  3. Judgement. They will have been asked to make many judgements in their work and had to defend them in discussions with their supervisors and fellow students. Unconvincing opinions, not backed by evidence, will not pass muster.
  4. Cognitive sovereignty. Students will have to take their own stand on their work, after all of the challenge and argument. Independence of thought is the hard-earned outcome here.

And then they will have a fighting chance of being centaurs in the world awaiting them, masters of the technology and opportunities available, rather than the reverse-centaurs that our education system is currently, and inadvertently, preparing them to be.

The AI Rollercoaster

11. June 2026 · Write a comment · Categories: Actuarial, AI, COVID-19, Education, Professions
Source: https://xkcd.com/249/ This work is licensed under a Creative Commons Attribution-NonCommercial 2.5 License

It all started for me in May 2017, with the challenge from Daniel and Richard Susskind in their book “The Future of the Professions”, which set out two possible futures for the professions. Either:

  • They carry on much as they have since the mid 19th century, but with the use of technology to streamline and optimise the way they work; or
  • Increasingly capable machines will displace the work of current professionals.

Their research suggested that, while these two futures would exist in parallel for some time, in the long run the second future would dominate. The actuarial profession was going to be particularly vulnerable. As the Susskinds wrote:

Accountants and consultants, for example, are particularly effective at encroaching on the business of lawyers and actuaries.

I stood in the Institute and Faculty of Actuaries Council elections that month, on a platform saying that we needed to urgently respond to this challenge. I didn’t get elected.

However, at the University of Leicester we pursued a curriculum transformation programme in response to this challenge aimed at developing actuaries of the future who had:

  • Highly developed presentation skills, both in writing and in speech
  • Great team working skills
  • Strong IT skills – comfortable with working with data
  • Clarity about why they are there and the desire to use their skills to solve problems

The 2017 post also talked about emerging trends which had hardly started at all yet:

  • The end of reserved roles for actuaries
  • Different ways of communicating advice
  • Online self-help for users of actuarial advice
  • The advance of roboactuaries and their assistants

A paper called 2036: An actuarial odyssey with AI written for the Society of Actuaries in July 2016 by Dodzi Attimu and Bryon Robidoux discussed the possibilities for robo actuaries and robo actuarial analysts.

They estimated that robo actuarial analysts would be with us in 5-10 years and would provide:

A system that has limited cognitive abilities but can undertake specialized activities, e.g. perform the heavy lifting in model building (once the specification/configuration is created), perform portfolio optimization, generate reports including narratives (e.g. memos) based on data analysis, etc.

Whereas a robo actuary was more like 15-20 years away and which they also helpfully described:

We mean a software system that can more or less autonomously perform the following activities: develop products, set assumptions, build models based on product and general risk specifications, develop and recommend investment and hedging strategies, generate memos to senior management, etc.

Then, in 2018, there was the whole Bullshit Jobs argument posed by the late David Graeber, discussing Keynes’ prediction in 1930 that:

In quite a few years – in our own lifetimes I mean – we may be able to perform all the operations of agriculture, mining, and manufacture with a quarter of the human effort to which we have been accustomed.

Graeber, in Bullshit Jobs, pointed out that this never happened, despite pretty much all of the technological developments and income increases which Keynes predicted. He suggested that the future which the Susskinds were predicting is already happening in terms of needing fewer people to fill the meaningful roles within organisations but that, rather than employing fewer people, we are either creating “bullshit” jobs which even the people doing them can see no point to or bullshitizing existing roles for which the meaningful need has passed. It was as if the organisations themselves have attempted to maintain the outward appearance of the same structures by disguising the hollowing out of so many of their functions with simulated business.

Why? One of the reasons he thought the situation had been allowed to develop was that noone believed that capitalism could produce such an outcome. Graeber gave the example of the creation of Obamacare, where Barack Obama “bucked the preferences of the electorate and insisted on maintaining a private, for-profit health insurance system in America” in order to protect jobs in health insurance.

Then we had the pandemic, and the painful return to work which found that people were not necessarily keen to return to populating those office empires, preferring to work remotely. Some of the attempts by the captains of industry to get them back were a little desperate.

Now nine or ten years on from the initial challenge, we are deluged in articles about how AI is impacting different areas of actuarial work, whether it is already replacing graduate roles and what actuarial students need to do to make themselves employable. And now the blinkers also seem to have come off about capitalism not producing the need for fewer jobs.

Ian Pay of the ICAEW’s quote from last year was just one example:

Historically, accountancy firms have typically had a pyramid structure – wide base, heavy graduate recruitment. Firms are now starting to talk about a ‘diamond model’ with a wide middle tier of management because, ultimately, AI is not sophisticated enough yet to make those judgment calls.

But hang on a moment. Now there is something called an “AI boomerang”. Sam Altman of Open AI and Dario Amodei of Anthropic have both been backtracking on their predictions of job lay offs due to AI. As the Gold and Geopolitics Substack puts it:

Two-thirds of the companies that ran AI-driven layoffs last year are now rehiring. One in three spent more on the rehiring than they saved on the original cuts. Robert Half calls them “AI boomerangs” – which is the name a consultant invents when there’s a market in unwinding what the last consultant sold you.

The last few years of AI-generated headlines (in both senses of the phrase) have been quite the rollercoaster.

I do not want to add to the deluge, as currently you would need a LLM summarising reports for you 24/7 just to keep up with it as it is.

And there are good reasons for not rushing to judgement on this. It is only six weeks ago that the Bank of England’s Head of Financial Stability warned that global stock markets are overvalued. As she said:

The thing that really keeps me awake at night is the likelihood of a number of risks crystallising at the same time – a major macroeconomic shock, confidence in private credit goes, AI and other risky valuations readjust – what happens in that environment and are we prepared for it?

Since then, the economics of the major AI players has only got more bonkers. In this environment, there is considerable uncertainty about what students should be learning to prepare them for the world they will need to rebuild from the rubble of the current one.

Daniel Susskind reckons a “no regret” strategy for education would focus on the basics, which he describes as literacy, numeracy and critical thinking, and the critical use of AI. Carlo Iacono talks about the work which will always be hard to automate as follows:

The work that depends on reading a room. The work that relies on institutional memory. The work where the important fact is not in the document. The work where the answer is political, ethical or relational as much as technical. The work where being right is not enough, because someone has to be accountable for the consequences.

But you have to get into the organisation first before you can develop much of this. I remain reasonably comfortable with our prescription from 2017 in terms of broadly what the curriculum should be trying to achieve. Curriculum changes take time and, tempting though it is to pile into syllabus changes aimed at incorporating the latest cutting edge technological developments, the likelihood is that you will be arriving at the wrong fire. As a social media post this week put it:

The danger is that, as Agentic AI in particular looks so much like the predictions of 2016, we declare it the new messiah and bet everything on this being the future. I certainly think that 2036 will look very different to now, but I am not convinced that we have the shape of it yet. This may just be another bullshit alternative, waiting for the next crisis to mutate into something else.

What we can say with certainty is that the future for early professionals is as uncertain as any of us can remember. And, in my view, the best way to support our young people starting their professional lives (or whatever is going to replace professional lives) is to make the value they add much clearer to anyone who might want to work with them in the future.

Future education systems are going to need to help their graduates demonstrate their value in ways they haven’t historically needed to. AI has not brought new problems, it has accelerated existing ones. And the gap between actual actionable skills, knowledge and experience and the credentials which are supposed to represent them is the key one I believe as far as those graduates are concerned.

Because currently the companies who might want to take them on have to guess what they can do to a large extent. What does a 1st mean? A 2.1? Do the differences matter to their employer? Should they interrogate the details of student courses? How would they incorporate that into a manageable recruitment process? Some people see a future of AI generated applications by the thousand per student doing battle with AI powered triage systems operated by potential employers.

This is already an issue in academia. The conclusions of one academic paper on submissions to one major academic journal were stark:

Submission volume has risen 42% since the late 2022 release of ChatGPT, while writing quality has declined. The rise in AI-generated writing accounts for nearly all of these trends.

On the other hand, researchers complain about the increasing use of AI peer review to try and cope with these increased volumes, and some have tried to game the LLMs they believe they are dealing with.

I think we can do better than this in actuarial teaching and learning.

And this will be the subject of my next post.

Answering a better question

05. June 2026 · Write a comment · Categories: Actuarial, AI, Economics, Education, Professions
Source: https://xkcd.com/1319 This work is licensed under a Creative Commons Attribution-NonCommercial 2.5 License

I didn’t understand statistics until I started taking actuarial exams that required me to master particular statistical techniques, decide which ones were appropriate to the problem I was looking at, apply them accordingly and be able to interpret what the results did and didn’t tell me. An A at GCE O level in maths, and A at GCE A level in both maths and further maths and a maths degree from Oxford did not give me those abilities.

I didn’t understand economics until I started putting together modules which could be taught on both BSc and MSc courses and then teaching them. My economics module on the way to qualifying as an actuary, based very heavily on Economics by Begg, Fischer and Dornbusch, did not give me that understanding.

There is a pattern here: the most impactful experiences we have are frequently at a bit of a distance from the credentials we present to the world. Your education is not a paragraph on your CV, it is your lived experience, sometimes assisted by, sometimes actively hindered by and often pursued completely independently of the educational institutions you have had a relationship with during your life.

Daniel Susskind has marched into this often fraught relationship between credentials and actual actionable skills, knowledge and experience in the last lecture of his Gresham College series on The Future of Work, called Education – And Its Limits. His analysis is a very clear expression of the problems that will be created if AI systems prove to be half as capable and long-lasting as people from OpenAI and Anthropic are telling us they will be.

Susskind argues that trying to future-proof students through education was a hopeless task and that working on the assumption of unresolvable uncertainty was the better way forward. He suggests a “no regret” strategy for education would focus on the basics, which he describes as literacy, numeracy and critical thinking, and the critical use of AI. Others in the audience suggested some other “basics”: communication skills for instance.

The other strand of Susskind’s basics was critical use of AI. And his challenge to the audience was whether we can teach AI without losing the basics in the process.

After a bit about the need to make continuing education in later life more accessible, he looked at his list from a previous lecture (which I briefly touched on here) of problems for a post AI future:

  1. Distribution (replacing wages);
  2. Contribution (how do you “pull your weight”);
  3. Power (domination by Big Tech on economics, politics, liberty, social justice and democracy); and
  4. Meaning (fulfilment in life).

And this is why I continue to watch Susskind’s output, because he is the unusual combination of an extremely orthodox economist (see his book Growth: A Reckoning for proof of this) and someone who has been wrestling with the challenges of more capable systems to our way of doing things for over 10 years. What Susskind gives you is a peek at how the actual economists advising our governments would deal with things if OpenAI and Anthropic are proved right. It is as if you had someone who both thought, as Ben Bernanke, Chair of the Federal Reserve, did in October 2007, that “the banking system is healthy” and also that the banking collapse was going to happen anyway.

Because what he reveals is that, if Anthropic are right, orthodox economists have really got no policy prescriptions worthy of the name.

On 1. Distribution: Susskind acknowledges that, if the labour market could not redistribute wealth effectively any more, then a bigger state would be needed to do so (but he was at pains to emphasise that this would not be the 20th century central planning type of state).

On 2. Contribution: Susskind thinks that perhaps we should allow people to make non-economic contributions! As if all of the activity in society which economists routinely ignore really wasn’t already happening!

On 3. Power: Susskind says the political power of Big Tech with regard to liberty, social justice and democracy is a problem. We have anti-trust legislation that can deal with Big Tech’s economic power, but not its political power. I don’t know what kind of political power he thinks Big Tech would have without the economic power that has been granted it by a steady erosion of that anti-trust legislation over recent years.

On 4. Meaning: he has little to say other than something about us currently having policies for work but not for leisure.

And in all of this, there is the implicit underlying assumption of a stable future environment for all of this tech to operate within.

For me it brought to mind something a friend of mine who was in a tent with Cory Doctorow at the How The Light Gets In festival at Hay-on-Wye last week. “He has a theory of change” he said.

He really does, including about how to break the economic power of the Big Tech companies. You all need to read Enshittification for the full account, but my review of the book here is a sneaky peek.

Susskind really really does not have a theory of change. Which tells me that the economics profession does not have one either.

However I continue to watch him as I find he goads me into thinking what some better answers might be to the questions he asks. And perhaps also a better question than whether we can teach AI without losing the basics in the process. I think a better question would be what do we need to learn when the future is uncertain.

First of all, let’s remind ourselves of the problem. If noone cares how your advice was constructed, but your client can get advice that ticks the compliance box more cheaply and quickly from an AI system, while the experienced professional still has some role in managing the process, it may increasingly be a struggle to justify the cost of the junior colleague. So the future education system is going to need to help that future junior colleague demonstrate their value in ways they haven’t historically needed to. AI has not brought new problems, it has accelerated existing ones. And the gap between actual actionable skills, knowledge and experience and the credentials which are supposed to represent them is currently the key one as far as that future junior colleague is concerned.

The temptation is to rush into syllabus changes towards what currently looks like the cutting edge activity. I agree with Susskind here that this would be a mistake. He cites the example of Michael Gove, amongst many other education ministers at the time, mandating the teaching of coding in 2014. Now we find that the new AI systems (despite the problems highlighted by Hannah Fry, Kyle Kingsbury and others I talked about here) are most suited to writing code and Anthropic claim that Claude is now writing 80% of its own code. Programmers are saying that, on the famous XKCD cartoon above, they are now living on the theory curve.

But both higher education institutions and the professions who still want to be in the game of developing the next generation of professionals can do a lot more both to reduce the gap between credentials and actual actionable skills, knowledge and experience and to make it clearer to employers that they have done so. That will be the subject of my next post.

« Older articles