A woman moves through a quietly intelligent morning kitchen at golden hour, illustrating AI predictions about invisible infrastructure seamlessly embedded in everyday life."

#65 | My six AI predictions for 2026

ByMark Updated:

TL;DR: Six AI predictions for 2026 — each one grounded in research, each one carrying a complication the headline version usually skips. Here’s what’s already closer to your daily life than you might expect.

👋 Happy Friday,

The month of January has its rituals. Resolutions, gym memberships, and somewhere in between, a surge of AI predictions.

Each of those AI predictions arrives confident and fully formed, knowing exactly where the next twelve months are headed.

But confidence is easy to produce. Specificity is a bit harder.

A prediction that names something already underway, something with real evidence sitting behind it, tends to feel less exciting or dramatic than one promising a revolution.

That might be true, but it’s at least worth your time to get an idea of what might and could be coming.

I have chosen six of those less exciting and boring predictions for our edition.

Each comes from researchers and institutions tracking this closely. Each carries a complication the headline version usually skips.

And somewhere in each one is something already touching the way you work, communicate, or move through an ordinary day.

The crystal ball is on. 👇

The AI Learning Guy newsletter 🤖 🧠💡

AI learning hacks and mega prompts delivered to your inbox.

1. AI will become invisible infrastructure

MIT Technology Review’s senior editors said it plainly earlier this year: AI will fade into the background and become embedded in everyday workflows.

The sign that it worked, they said, will be that it disappears.

That’s a strange definition of success.

Think about electricity. Your refrigerator runs on it. Your phone, your lights, the heating that kicks in before you wake up.

Nobody thinks about electricity anymore — not because it stopped mattering, but because it worked well enough to stop being noticed.

You remember it exists exactly once, when it stops.

That’s the trajectory MIT is describing. Not AI as something you actively turn on and off, but AI quietly working behind the scenes in the things you already do.

For example, filtering your inbox, shaping your search results, informing your medical records, and routing your deliveries. It’s actually already present and making decisions. Just not loudly.

This is less dramatic than a revolution and harder to argue with than a forecast.

The complication is sitting right inside the prediction. Invisible things are difficult to question.

Once something becomes infrastructure, the habit of asking what it’s doing tends to go with it.

Which makes the next prediction worth reading carefully. Because invisible infrastructure that starts making decisions on your behalf is a different thing entirely.

2. Agentic AI is the industry’s biggest bet for 2026

An AI agent doesn’t just answer questions. It works independently, planning and executing tasks over extended periods without oversight.

Every major analyst is pointing to 2026 as the year this either proves itself or quietly stalls.

A CrewAI survey of 500 senior executives, published in February, found that 100% of enterprises plan to expand their adoption of agentic AI this year.

A separate figure, from the same research landscape, shows that only 14.4% of teams deploying these agents have received full security approval.

So ambition runs well ahead of readiness.

The systems making autonomous decisions on behalf of companies are largely operating outside approved governance frameworks. Nobody announced this. It’s just how the numbers settled.

This adds another layer to what prediction 1 described.

Infrastructure becoming invisible is one thing. Infrastructure becoming invisible while acting autonomously before the rules around it are written is where it gets truly interesting.

3. LLMs are contributing to verified scientific discoveries

DeepMind’s Gemini Deep Think contributed to research in mathematics, physics, and computer science earlier this year.

One paper was accepted at ICLR 2026. Several more are currently under review by journals. OpenAI reported that GPT-5.2 contributed solutions to previously unsolved mathematical problems.

However, these are peer-reviewed results, not press releases.

Researchers still direct the models. Scientists set the questions, evaluate the outputs, and decide what matters.

AI then explores the space faster than any human team could manage on its own. The discoveries are real, although the autonomy isn’t there yet.

Something has crossed a threshold, though. Because so far, science has always moved at the speed of human attention.

Attention is now being extended by something that doesn’t lose focus or accumulate fatigue across a long research cycle.

Where that leads for medicine, climate research, or materials science remains genuinely open.

For now, the prediction is simply that this accelerates in 2026. And, the evidence suggests it already has.

Which raises a question sitting just underneath prediction 4.

If AI is lowering the barrier to scientific discovery, what happens when it starts lowering the barrier to building software?

4. Software is increasingly being built by people who cannot code

Andrej Karpathy coined the term “vibe coding” in early 2025.

You describe what you want in plain language, and AI builds it. No syntax required, no programming background assumed.

Collins Dictionary named it Word of the Year. That alone is worth a moment.

By 2026, 41% of code globally is AI-generated, up from 8% in 2023. Ninety percent of developers use AI coding assistants daily.

And perhaps more telling: 63% of people currently using vibe coding platforms identify as non-developers.

Consider what that means for someone running a small business, teaching a class, or managing a team.

Tools that previously required hiring a developer can now be sketched out in a conversation. The barrier that kept software creation inside a specific profession is dissolving.

What sits underneath that tool, how it was built, and what it assumes, remains largely invisible to the person using it.

Useful and unexamined tends to be a familiar combination, as prediction 1 already suggested.

5. AI is quietly entering homes, relationships, and daily routines

Bryan McCann, co-founder of You.com, predicts household robots will become normalised in 2026 and will be available on monthly payment plans.

That timeline remains to be seen. But what’s already underway arrived with little fanfare.

AI manages grocery orders, plans travel, and handles smart home interactions for an increasing number of people.

Voice-first interfaces are replacing typing as the primary way people interact with AI systems. Rather than searching or clicking, people are increasingly just talking.

The American Psychological Association began formally tracking AI companions and digital relationships in 2026 as a monitored psychological trend.

Not as a distant concern. As something already present enough to measure.

Pilot programmes in nursing homes found that AI companions reduced behavioural problems and lowered psychotropic medication use. For isolated elderly people, that result is meaningful.

For people with pre-existing attachment difficulties, research points elsewhere. Extended use of AI companions appears to deepen loneliness rather than ease it.

In a nutshell, the same technology produces opposite outcomes, depending entirely on who is sitting with it.

6. A measurable part of the population is choosing to go offline

Somewhere alongside the adoption curve, something quieter has been building.

Futurist Sinead Bovell began documenting the signals in late 2025.

Social media usage is declining. People are shifting from public feeds to private messaging. Running clubs are gaining members while dating apps are losing them.

Communities choosing proximity over algorithm.

Teenagers, without parental mandates, are voluntarily reducing their own screen time. Research shows they recognise the effects on sleep and concentration and are adjusting on their own terms.

Richard Socher, co-founder of You.com, predicts this will harden into multiple waves of people who simply decide that things are good enough as they are.

Two things are accelerating simultaneously in 2026. AI is increasingly embedded in the texture of daily life.

A growing number of people are actively stepping back from it. These aren’t contradictions. They’re the same moment, experienced from different distances.

Both of which, in their own way, are worth paying attention to.

Alright, that’s it for today.

Six predictions, each one already in motion. None of them finished arriving.

Consequential changes rarely announce themselves clearly.

They tend to show up quietly, in the background of an ordinary week, until one day you simply can’t remember things being any other way (think of smartphones, the internet, social media).

Cheers,

Mark
The AI Learning Guy
👋⚡😎

The AI Learning Guy newsletter 🤖 🧠💡

AI learning hacks and mega prompts delivered to your inbox.

Interesting Sources

Note: No single website has all the answers. This list serves as a starting point for those who want to explore or satisfy their curiosity about AI.
Links: Links with * are affiliate links. See disclosure below.