Chicago AI Meetup Reaction (Hype Cycle Litmus?)

If you’re looking for a litmus test for where we are on the AI Hype Cycle curve, find a local AI meetup.

A good one might be AICamp; it’s a syndicated AI meetup held in most major cities.

Tonight I visited the Chicago AI Meetup at the Chase Building downtown. I’m new to the city so I had to see what all the fuss is about.

Here’s where we’re at in the hype cycle:

I walk in the lobby about 40min early for the event and was met with a line of maybe 100 people already waiting.

Immediately any outsider can tell we’re all here for the AI meetup... we’re an incredibly diverse group that somehow all look the same.

The line eventually moves and I get to the registration table. They need to see my Driver’s License and my event confirmation email to let me in.

I find out once I’m past the threshold that nearly 500 people RSVP’d for tonight’s event… a new all-time high.

I grab the last of the pre-meeting snack assortment and join a table of a few other AI heads. They’re all college students, two of them are already running $5K MRR startups. We make small chat by comparing AI thoughts and LinkedIn profiles.

When the time comes we all funnel into a big auditorium for the tonight’s speakers, it' starts with a round of applause, there’s a buzz in the room.

An admin shares some recent events and a QR code to join the group’s Discord. A few people announce projects they are working on or looking to be a part of, then the speakers come on.

It’s 6pm. Tonight there’s 3 speakers from big giant companies, each has 30 min to share some research. Google, IBM, and JPMorgan.

First speaker shares a bit about how JPMorgan is building their own internal LLM Suite and Document system, which they’ve already shipped to 150,000 of their employees

Second speaker is from Google. They showcase the state-of-the-art LLM’s and the history of algorithms that preceded, plus a nice overview of RAGs and LangGraph.

The third speaker is a fill-in from IBM. He’s older. He talks about how InstructLab can be used to incrementally train LLM’s for enterprise use cases using a combination of student + teacher + critic models, and presents some ideas about how to design robust testing frameworks for modern LLMs (a notoriously pesky problem is automated testing of probabalistic frameworks).

All three speaker stick around for a panel of moderated questions, which range from ‘how did you get into AI’ to ‘what comes next for LLMs’?

They talk through ideas for merging / combining open-source models (no good way to do this nowadays), share ideas for how agent-based models will evolve, and reassure the crowd that as great as AI has become for solving mundane productivity-related tasks, we are very far away from anything resembling super-intelligence.

By now it’s 8:45pm, the crowd has thinned out but a faithful group of about 90 remain, peppering the panel with questions. Young students and grads starting AI businesses, older people with legal pads taking notes and recording slides on iPhone 5’s, everywhere in between.

I leave by 9pm to get back home, the last of the contingent is still buzzing throughout the corridor.

So if you were to ask me where we are in the AI hype cycle, I’d say 1) the hype is as high as it’s been, and 2) there’s room yet for it to keep getting higher.

Walking back home I’m struck by a thought — if this is how it is here in Chicago… what the hell is it like in The Bay area these days?