Simply while you suppose you’ve wrapped your thoughts round computer systems that may put your dog in front of the Eiffel Tower or chatbots that act like your best friend (or lover), the AI behemoths shock you with a completely AI-powered TikTok or the power to virtually bring back your dead relatives.
I’ve labored within the AI area for 15 years. I served as an early beta tester for OpenAI in 2020, once I predicted that a little model called GPT-3 had world-changing potential.
It was later launched as one thing referred to as “ChatGPT”–maybe you’ve heard of it?
I’ve additionally called several big AI trends correctly, together with the rise of video mills and the “AI Wars” between Google and OpenAI.
Based mostly on my expertise, listed here are my six AI predictions for 2026 and past.
1. OpenAI goes garlic
In late 2025, Google’s Gemini mannequin started to gain ground on OpenAI and its GPT-5.1 system.
That apparently actually irked Sam Altman and the OpenAI staff. Altman reportedly referred to as a “code purple,” directing workers to focus all their efforts on besting Google.
Rumor had it they have been growing a brand new, totally re-trained considering mannequin, codenamed Garlic. When OpenAI did a shock drop of a brand new GPT-5.2 mannequin in mid December, plenty of folks thought it is perhaps Garlic coming to market.
Based mostly by myself testing, it’s not. Or not less than, it’s not the entire mannequin.
GPT-5.2 is certainly higher than the earlier mannequin. It’s quicker and extra environment friendly, and makes fewer errors. It’s additionally notably higher at scientific duties, and sensible ones like coding.
However now it seems there shall be a second new OpenAI launch, which I count on to return out in January. That’s probably the complete Garlic mannequin.
This new mannequin, I predict, can have a brand new data cutoff someday in 2025, a broader context window, and significantly better picture era capabilities. It is going to even be quicker and extra environment friendly to run, particularly on “considering” duties.
2. Google’s Gemini continues its march towards domination
Each time it lastly arrives, Garlic will enter the world with loads of competitors.
Google was very sluggish to roll as much as the generative AI desk. For an organization that’s been working in deep studying for many years and has among the most clever folks on the planet working for it, that felt like an enormous miss.
Google had reportedly developed its own ChatGPT years earlier than OpenAI, however chickened out on releasing it. At first of the AI race, that allowed OpenAI to very loudly and publicly eat Google’s lunch.
The historical past of science, although, is suffering from examples the place early innovators weren’t those who efficiently commercialized a brand new expertise.
Simply ask Joseph Swan, the true inventor of the lightbulb. You’ve by no means heard of him. However you do know Thomas Edison, who made the lightbulb a broadly out there expertise—and did an important job selling his invention (and himself) within the course of.
Traditionally, first-mover benefit has confirmed surprisingly inconsequential within the tech area. And now that Google has woken as much as the significance of AI, they’re aggressively constructing out their Gemini mannequin and integrating it into virtually all their merchandise, together with their core search expertise.
Google has extra knowledge, extra assets (together with its personal customized AI chips), extra folks, and a wider attain than OpenAI.
In 2026, Google will proceed to throw its weight round in generative AI. Gemini will go from being an also-ran to some of the highly effective fashions available on the market. As a result of will probably be broadly built-in into merchandise that ordinary folks use on a day-to-day foundation, it would instantly have an viewers within the billions.
The wrestle now isn’t for newer corporations like OpenAI to create the perfect product. It’s to create a product higher than Google’s. That shall be very exhausting in 2026 and past.
3. Chatbots turn into remedy (and a bit extra)
Customers have already realized that ChatGPT can take the place of a human therapist. In a recent poll by the Economist, 25% of individuals reported turning to chatbots for psychological well being help.
As circumstances of AI psychosis and alleged suicides demonstrate, this may go very badly. However for individuals who can’t afford any form of psychological help—or just don’t have entry to it of their language or nation—utilizing chatbots as low-cost therapists is extremely interesting.
With out instantly saying so, OpenAI has implied that they’re transferring ChatGPT additional into this area, with improvements to how the bot handles sensitive medical and mental health conversations.
This might be an enormous boon for psychological well being. Many individuals seem extra comfy discussing their issues with an unthinking bot than with a human. The truth that a session with ChatGPT doesn’t value $300 per hour can be an enormous plus!
ChatGPT’s capabilities will increase in different methods, too. A rumored “grownup mode” will arrive in 2026, permitting ChatGPT to jot down risqué materials. Put together your self for a wave of frenzied op-eds about how persons are turning to those newly salacious for relationships as a substitute of fellow people.
4. AI-generated movies take over—and never simply on Sora
OpenAI’s AI-powered Sora video generator is extremely highly effective, and their Sora-based social network is incredibly fun to use.
In 2026 and past, count on to see the attain and significance of AI-generated vertical video speed up.
Vertical video is the proper format for AI. The clips are typically brief, which caters to AI’s skill to generate about 10 seconds of dependable video earlier than issues go off the rails.
In addition they are typically grabby and compelling. Once more, AI excels at making videos of things like people falling into wedding cakes or having heated arguments with their roommates.
In 2026, count on the attain of AI-powered social networks like Sora to develop dramatically.
The most important progress, although, will come from these movies migrating off the Sora platform and onto different social media. Already, my Fb Reels feed is dominated by clearly AI-generated movies of issues like a cat saving her kittens from a flood or grandmothers preventing grizzly bears.
In 2026, AI movies received’t keep put. They’ll journey into each vertical video area on the internet—from TikTok to Nextdoor—additional blurring the strains between what’s actual and what’s imagined.
5. Electrical energy turns into the limiting issue
I’ve pals who construct knowledge facilities for a dwelling. They inform me the one factor stopping them from constructing extra knowledge facilities is discovering sufficient electrical energy to maintain up with AI’s calls for.
Some corporations are reportedly even going nuclear, constructing or recommissioning totally purposeful reactors to energy their electricity-hungry AI chips.
The necessity for extra electrical energy for AI will begin to restrict the tech’s progress in 2026. It is going to additionally rub up towards society’s different wants.
In 2026, I count on a populist backlash towards the truth that knowledge facilities’ voracious power calls for are elevating electrical energy charges for on a regular basis folks.
In the end, the deficiencies of immediately’s grid–strained as it’s by the rise of AI–will drive progressive, new fashions which are good for everyone. Low-cost solar energy at noon could also be redirected towards knowledge facilities, for instance, or saved in large batteries to maintain servers working in a single day.
This demand will create an enormous marketplace for inexperienced applied sciences, finally benefiting the planet and everybody on it.
6. AI invades the true world
No, the robotic rebellion isn’t right here simply but. However AI is more and more invading the true world.
Self-driving vehicles have been as soon as a novelty. In 2026, utilization will explode, with Zoox, Waymo and their rivals–together with entrants from China–serving extra cities.
The rise of self-driving vehicles and different bodily manifestations of AI expertise will shock folks. You’ll blink, and in the future it would really feel like practically each automobile on the highway is self-driving—because it at the moment does in my residence metropolis of San Francisco.
I count on to see different experiments with bodily AI in 2026, from robotic baristas to caregiving machines, and loads of army AI tech, too.
Once more, although, the self-driving automobile blitz coming in 2026 would be the most profound and shocking (to on a regular basis folks, anyway) implementation of the expertise. It is going to arrive far earlier than you suppose.
A pat on the again
In order that’s what I anticipate for the 12 months forward. As somebody who’s been within the AI area for a very long time, I’ve missed some issues.
However I additionally wrote, two years earlier than ChatGPT, that “OpenAI and its founders may simply make billions (and certain problem the promoting and content material suggestion engines of rivals like Google) by throwing warning to the wind and throwing open the doorways to GPT-3 to all comers.”
I nonetheless pat myself on the again for that one.
How will my 2026 predictions form up? Ask me in a 12 months!
