I tried to predict 2025 from 2015, and 2026 has marked my paper

At the end of 2015 I sat down and wrote what I thought 2025 would hold. It is 2026 now, a year past the line, so it is only fair I mark my own paper. Some of it makes me wince. Some of it I would write again without changing a word. Here is what I said about artificial intelligence at the time.

Artificial intelligence has made significant leaps and bounds in 2015 mainly fuelled by the cheap abundance of computing resource, miniaturisation of devices delivering more power and soon the introduction of commercial Quantum computing. Big data and analytics have also been significant investment growth opportunities couple with machine learning it means 2015 has left us with the building blocks for advancement in human history like no other time since the industrial revolution and electrification of industries.

The quantum part was early. The rest landed. That cheap abundance of compute is the whole reason a thirty-person firm in 2026 can put one curious member of staff in front of a tool like Claude Code on the 4.8 model and get work out of it that used to need a team and a budget. Then I turned to jobs.

In 2025 what we will see is a destruction of any industry that relies on a repetitive task such as accountancy or medical AI will be the 'brains' behind all our devices all of our choices all of our health services and above all our social views. The issue with such a thing is that we believe everything and trust whatever is 'on our screens' as humans we will need to use our intellectual capacity to move beyond 'acceptance' to challenging.

I had the direction right and the tone wrong. The repetitive task did not destroy the accountant; it freed the good ones. The warning I tacked on the end, about trusting whatever is on our screens, turned out to be the whole game. I was bolder still about education.

Is this the start of the Humanification we blogged about in 2013? How will our children learn in 2025? Will we see today's education methods dismissed as archaic? Are we looking at the rise of education algorithms that teach according to the subject's capacity to learn? An example I use from 2015 is the question of should a child need to learn maths to an advanced degree when calculators and a plethora of devices already perform this function?

On that one the jury is still out, though every teacher I know is asking a sharper version of the same question this year. On the home, I was confident.

We already have robotic hoovers, robotic industrial workers and a robotic seal that comforts the elderly in japan (yes really). As Artificial Intelligence increases we will see the use of more 'learning' capability and automation and freedom of robotic 'services'. Note I said 'services', it is my view that repetitive tasks such as domestic cleaning will be 'service based' the same way we order a taxi now.

Swap domestic cleaning for first-draft code, first-pass legal review, first-cut analysis, and that is 2026. We order intelligence now the way we used to order a taxi. On the fears, I reached for an old comparison.

The military forces of this world will always take technologies and abuse them for nefarious purposes. My view on the malevolent predictions of AI is that we have had nuclear weapons ever since 1946 and these have been improved on drastically since that time, have we seen a nuclear holocaust?

The part I missed, then and now, is the people. We did this with cloud. Firms bought the platform, moved the work, ticked the box, then found almost nobody on staff knew how to run the new thing well. So they panic-hired and let everyone else feel slow. Reskilling for AI is not a half-day course on writing a prompt; you give people real problems and real time on them, because your sharpest database person might be average at this for six months and then, one Tuesday, very good. The big firms, meanwhile, are asking the cloud-or-keep-it question again, this time about AIOps and the incident data a model would have to read. The honest answer is the dull one. It depends on what the data is and who is allowed to see it.

Back then I described the house itself.

Therefore every device we own will be 'smart' from the TV we watch to the earpiece we wear. Our interaction will be seamless thought controlled and tuned to our every whim and need by smart algorithms that run our lives.

But there was one passage I would not change a single word of.

AI and its mechanical cousin robotics to be truly aligned to human capability requires the one thing that during this Christmas time we all possess 'Humanity'. Humanity, thoughts, feelings and emotions will never be replicated until humans become a 'hybrid' with machines.

Eleven years on, that still feels like the heart of it. The tools will keep getting better; they are already better than most of us can use. I will end where I ended in 2015, with a line I have never improved on, from James Lovelock.

For each of our actions there are only consequences.

And with the note I left my future self, which I am keeping.

Maybe in 2025 I could look back at this article and laugh at its inaccuracies, remember though when cinema was first displayed people ran out of the theatre thinking the train was going to hit them.