Not long ago, the big question about artificial intelligence and professional careers was:

Will AI replace us?

It is an understandable question. But for actuaries, it may already be the wrong one.

In “The Future Actuary: Swipe Right on Curiosity, AI and What Comes Next”, I argued that the opportunity for actuaries is not to compete with AI. It is to combine actuarial judgement with AI capability.

Our latest market research gives us an opportunity to look at what is actually happening.

The Acumen Resources 2026 Actuarial Salary and Industry Insights Survey found that 83% of respondents use AI at work, compared with 47% in 2025 and 23% in 2024.

AI has moved from something actuaries were discussing to something many are actually using. So perhaps it is time to move the conversation on too.

“The question now is not simply whether actuaries will use AI. It is: what happens to actuarial work when they do?”

Jenny Johnston BAFS, FIA, FSAI, Director, Acumen Resources

From 23% to 83%: AI has moved quickly

Three numbers from our research tell the story:

Just as importantly, this is not solely a story about younger actuaries. AI usage was reported by 85% of actuaries within their first five years after qualification and 80% among those with 11 or more years PQE.

That challenges one of the assumptions I hear occasionally: “AI is really something for the younger generation.” Our results do not support that interpretation.

AI adoption is appearing across career stages. That makes the conversation much bigger than learning how to use a new piece of software. It raises questions about how actuarial roles themselves may change.

What are actuaries actually using AI for?

Among the 522 respondents who described how they use AI, the most commonly reported uses included:

Microsoft Copilot was the tool respondents mentioned most frequently.

Look at that list and something becomes apparent. Writing. Coding. Summarising. Research. Analysis. Automation.

These are not activities sitting somewhere outside actuarial work. They are activities sitting inside it.

Are AI skills already linked to actuarial salaries?

Our 2026 data provides one particularly interesting result. Actuaries who reported using AI were paid broadly in line with the market median for their level, while the 73 non-users in the comparison were 4.5% below the relevant market median.

That deserves attention. But it also deserves caution. Our results do not show that using AI causes someone to earn more. The respective populations may differ by role, employer or other factors. I would therefore be reluctant to describe this as an “AI salary premium”. It is not evidence of that.

Instead, it raises a more interesting question: could AI familiarity increasingly become part of the expected professional toolkit?

If 83% of respondents are already using AI at work, AI fluency may eventually become less of a differentiator and more of a baseline capability.

And if that happens, the differentiator moves again: from “Can you use AI?” to “Can you use it well, challenge it, apply professional judgement and turn it into a better business decision?”

AI is not necessarily replacing the job. It is changing the tasks inside it.

One of the central arguments in “The Future Actuary” is that we need to stop thinking about a job as one indivisible block. Every actuarial role is a bundle of tasks.

Some are repetitive and process-heavy. Some involve writing, documentation, coding and manipulating information. Others depend much more heavily on judgement, context, professional challenge, communication and influence.

The usage reported in our survey suggests AI is already touching several of the former categories.

If technology can help an actuary prepare a first draft, assist with code, summarise material or automate part of a process, the question should not automatically be how we preserve all of that work. It should be: what can the actuary spend more time doing as a result?

What becomes more valuable when AI does more of the production?

There is an understandable assumption that as machines become more capable, human professionals become less valuable. I do not think that follows automatically.

If AI becomes more involved in producing an output, someone still has to decide:

• Is it right?
• Does it make sense?
• What assumptions sit behind it?
• What is missing?
• What happens if the model is wrong?
• How should we explain the result?
• What decision should someone make because of it?

Those questions sound remarkably actuarial.

The opportunity is to move actuarial effort towards judgement, challenge, communication, insight, influence, ethics, context and decision-making.

That is not an argument for less technical expertise. Quite the opposite. You need sufficient expertise to recognise when an answer does not make sense.

The spreadsheet person needs a new profile

For a long time, technical expertise could become a large part of an actuary’s professional identity. You might be known as the Excel expert, the Prophet person, the reserving specialist or the one who understands a particular model.

Those skills can still matter. But if the tools change quickly, defining yourself primarily by the tool creates a problem.

“Tools age. Capabilities travel.”

Jenny Johnston BAFS, FIA, FSAI, Director, Acumen Resources

That is why career descriptions need to change too:

The person has not changed. The description of their value has.

That becomes increasingly relevant if AI starts performing more of the activities we previously used to define the professional.

AI could make human skills more visible, not less

The more technology can do, the easier it may become to recognise what the technology cannot provide on its own.

An actuary still needs to operate when evidence is incomplete. Someone needs to challenge assumptions, understand commercial context, communicate uncertainty and know when a theoretically correct answer is not practically useful. Somebody remains accountable for important professional decisions.

That is why I do not see the future profile as technical skills or human skills. Nor is it actuary or AI. The more useful combination is technical expertise + AI capability + human judgement.

There may also be an opportunity for experienced actuaries

The survey finding that 80% of respondents with 11 or more years PQE report using AI is especially interesting.

Experience and AI capability do not need to sit at opposite ends of the career spectrum.

An experienced actuary has something technology does not suddenly create overnight: professional experience accumulated through difficult decisions, incomplete information, unusual circumstances and outcomes that did not follow the model perfectly. In other words, judgement where the data runs out.

Pair that judgement with technological fluency and the result could be extremely powerful.

For an established actuary, the objective does not need to be competing with somebody who has been coding since childhood. It may be understanding the technology sufficiently well to combine it with what years of professional experience have already taught you.

Junior actuarial work may need to evolve too

Junior professionals have traditionally learned partly by doing some of the production work that AI may increasingly assist with: first drafts, repeatable analysis, research, documentation and coding.

Those activities do not merely create an output. They can also contribute to learning.

So if technology changes those tasks, employers should consider what professional development looks like alongside that change.

“The objective should not simply be ‘How much junior work can we automate?’ A better question is: ‘If the work changes, how do we make sure the learning still happens?’”

Jenny Johnston BAFS, FIA, FSAI, Director, Acumen Resources

The future actuary will not blindly trust the machine

There is one professional characteristic that becomes particularly important as AI adoption increases: scepticism. Not cynicism about technology, but professional scepticism about its outputs.

Whether the model is actuarial, statistical or AI-based, professional value does not come from accepting whatever appears on the screen. It comes from understanding it, testing it, challenging it and recognising when something does not make sense.

The future actuary is not someone who simply trusts the machine. The future actuary knows how to use it, question it and translate outputs into decisions.

AI literacy without judgement is not enough. And judgement without willingness to understand new tools may not be enough either.

What should actuaries do now?

I do not think everyone needs to become an AI specialist. What matters is engagement. Start small.

The objective is not to become more machine-like. It is to become more valuable with the machine.

Actuary + AI may be the real future career match

The 2026 Actuarial Salary and Industry Insights Survey captures a profession undergoing rapid technological change. Reported AI use has risen from 23% in 2024 to 83% in 2026, with respondents describing its use across writing, coding, summarisation, research, analysis and automation.

The implications will not be identical for every actuarial role. But the direction is significant enough that we can stop asking whether AI is relevant to the profession. It already is.

Now we need to decide what we do with it.

“The most interesting future is not actuary versus AI. It is actuary + AI.”

Jenny Johnston BAFS, FIA, FSAI, Director, Acumen Resources

Technical rigour combined with technological leverage. Experience combined with curiosity. AI capability combined with professional scepticism. Analytical power combined with the human ability to understand context, communicate uncertainty and influence decisions.

In “The Future Actuary”, I argued that tomorrow’s actuarial profile can be summed up as three things: Rigour. Range. Relationships.

Our 2026 AI findings make me more convinced that this is where the conversation needs to go.

“AI can change the work. Actuaries still get to shape what happens next.”

Jenny Johnston BAFS, FIA, FSAI, Director, Acumen Resources