10 Predictions On The Future Of AI After GPT-5 Failed
Time to start preparing for the great defunding
I’ve been saying for months that GPT-5 would be the turning point. Not because it would be great or AGI, but because it would expose the absurd gap between hype and reality. I have published my predictions on GPT-5 before it was released and was not surprised with the great disappointment:
And now that my predictions about GPT-5 came true and the disappointment is everywhere, I am ready to make another 10 predictions on what happens next and how to survive the AI winter if you are, like myself, in the field.
I have spent 17 years in NLP, I have seen hype and AI winter before but nothing to the extent of what is happening now.
So buckle up and let’s think what to do now when AI and everyone who is connected with it are losing trust thanks to a bunch of greedy tech bros with their unrealistic promises.
For two years we were told to wait for the savior. GPT-5 was supposed to be the model that would “deliver us from hallucinations.” The evangelists were preaching, investors were holding their breath, and OpenAI dragged out the release as long as possible to keep the myth alive. But when the model finally arrived, it didn’t deliver, it disappointed.
It showed that the king had no clothes.
, one of the people, who was the voice of sanity in the last years, got into the spotlight. Suddenly, the news started printing articles that [oh, there was one researcher who was right]. And at the same time, all the AI CEOs are suddenly hiding in bunkers; some are backpedaling [Altman sees AI bubble]. The influencers have stopped promising AGI breakthroughs and started posting anti-AI posts (search “AI bubble“ on LinkedIn and then look up the earlier posts of the OPs), as if they always knew, and did not write just a few months ago that coding, Google, white-collar jobs, or whatever is dead. Even Elon Musk, playing his eternal antagonist role as the archenemy of Sam Altman, is in trouble and is [suing everyone around.]They all built their stories on a promise of one model that didn’t live up to the fantasy. GPT-5 has become the moment that damaged not only OpenAI, but its competitors, the research community, and the entire field of AI. And this is exactly what the researches like Gary or myself were warning about. We were not AI haters, we knew that lies undermine trust. The AI evangelists will switch their narratives and will find a different field to revolutionize with their MBAs but those who dedicated their life to AI, like myself [source], will have to rebuild this trust. Remember: people were expecting nearly AGI-level performance [Altman “knew “how to build AGI] and just got another model indistinguishable from the latest releases.
And with that, here are my predictions for what happens next - and how to prepare for it.
1. The Great Defunding
Next year, you’ll struggle to secure budget for new POCs. Your AI Center of Excellence may lose headcount. When companies cut costs, AI spending will be the first thing under the microscope.
There were billions invested and only 5% of organizations saw the ROI according to an MIT report [GenAI Fails to Pay Off]. There are also other signs of waning investor trust: concerns about overvalued AI stocks, the looming threat of tighter regulation [source]. Already at the end of 2024, [Economist] pointed out that most of the value comes from Shadow AI, where employees quietly use AI to speed up their tasks. But as the MIT report made clear, usage hasn’t translated into profit and loss improvements. Or, as I like to put it: writing emails faster doesn’t make money for a company.
All in all, it’s likely that next year management will demand hard, measurable KPIs, not vague promises of productivity or broad “usage.” We’ve already established that usage does not directly translate into profit. The fact that many employees use ChatGPT to write meeting summaries doesn’t mean the company saves millions.
If you want to survive this wave and keep your budget, build your evidence now. Don’t bring vague claims like “we write code faster.” Bring hard KPIs:
Dollars saved
Time-to-market reduced
Customer satisfaction improved
So when the next budget meeting comes, you can stand up and say: “Wait, we’re the 5% who actually got ROI from GenAI.”
And if you didn’t? Then it’s time to reevaluate your strategy — which use cases you prioritize, whether to buy or build, and how well AI is integrated into internal processes. Otherwise, the defunding might be entirely justified.
Further in the article: predictions about the future of AI, such as the loss of strategic significance, the death of RAG chatbots, and why one should bet on AI-assisted coding along with recommendations on how to prepare now.
Paid subscribers get access to the full archive, quarterly roundtables, and priority individual chats to consult on AI topics.
Founding subscribers receive a 45-minute one-on-one call to discuss the AI topic of their choice—whether strategy, boosting developer productivity, identifying meaningful use cases, etc.—plus top-priority access in individual AI chats.
Also buy an item in airealist shop and get promoted to paid subscription for a month for each purchased item: