At COSM 2025 (November 19–21 in Scottsdale, AZ), Uli Homann, a Corporate Vice President at Microsoft, will address a pressing question: “The AI Revolution: Is It Living Up to the Hype?” — followed by a fireside chat with tech strategy planner Bob Kelly.

It is shaping up to be a good season for his talk. Just yesterday, news came out that Meta is laying off 600 employees from its AI department. At Axios, Ina Fried reports,

Between the lines: CEO Mark Zuckerberg grew concerned several months ago that the company’s existing AI efforts weren’t leading to needed breakthroughs or improved performance.

That conclusion led to this reorganization, the launch of TBD Labs, and the pricey hiring binge that coincided with Meta’s $15 billion investment in Scale AI and the hiring of [Meta chief AI officer Alexandr] Wang.

“I’m really excited about the models we’re training, our compute plans and the products we’re building, and I’m confident in our path to build towards superintelligence,” Wang said in the memo.

“Exclusive: Meta slashes jobs in its AI operations,” October 22, 2025

Wang is so excited that he’s slashing jobs? This and similar news gets spun different ways, depending who you talk to.

The limits of just getting bigger

Essentially, there is a rather large gap between what many industry boosters expected of new AI models and what they can really provide. A number of our writers — Brendan Dixon, Jeffrey Funk, Eric Holloway, Robert J. Marks, and Gary Smith — have documented this issue over the last few years.

First, there are limits to what can be achieved by making computers bigger and more powerful (the limits of scaling).

AI analyst Gary Marcus puts it succinctly, “Scaling has made GenAI much better at what it is good at, but not helped much with its core weaknesses around reliability. Hallucinations, poor reasoning, poor planning, etc, or its challenges in the face of novelty.”

That makes sense. Making a system bigger can’t fix weaknesses that its nature embodies. They just become weaknesses on a larger scale.

A classic example is ChatGPT telling economics professor Gary Smith last year that the Soviets had sent numerous bears into space, listing their names. Of course, no nation has sent any bears into space.

That error, Dr. Marks tells us, has since been found and fixed. But many other potential bloopers, built into the very way the system works, await hapless enquirers.

Overall, the the Singularity where we merge with machines, as predicted by Ray Kurzweil at COSM back in 2023, is just not happening. On the contrary, Andrej Karpathy, co-founder of Chatgpt’s maker, OpenAI, is dialing all the way back to reality:

The OpenAI cofounder, and de facto leader of the vibe-coding boom, appeared on the Dwarkesh Podcast last week to talk about how far we are from developing functional AI agents.

TL;DR — he’s not that impressed.

“They just don’t work. They don’t have enough intelligence, they’re not multimodal enough, they can’t do computer use and all this stuff,” Karpathy, who is now developing an AI native school at Eureka Labs, said. “They don’t have continual learning. You can’t just tell them something and they’ll remember it. They’re cognitively lacking and it’s just not working.”

“It will take about a decade to work through all of those issues,” he added. Lakshmi Varanasi, “OpenAI cofounder Andrej Karpathy says it will take a decade before AI agents actually work,”

Business Insider, October 19, 2025

AI claims now get sharper scrutiny… among those in the know

There are increasing complaints about workslop — fun but unproductive uses of AI that waste employees’ time.

In any event, firms are not rushing to adopt it; adoption may even be slowing down. Straw in the wind: That annoying background noise of over-hyped claims is being called out and even mocked at AI meets.

On the other hand…

Lots of people out there really think that AI models are just like humans: “People are falling in love with chatbots, speculating about whether they feel pain, and treating AI like a God reaching through the screen. There have been funerals for AI models and parties dedicated to debating what the world might look like after machines inherit the Earth.”

Some even want them to have personhood rights (Wired, September 4, 2025). Other folk worry that AI will wipe us out. But some may welcome that idea.

So where are we really with AI? Given all that’s happening out there, this is one panel you won’t want to miss. Here’s the agenda. The talk (and chat following) start at 1:15 on Thursday, November 20. Register here.