AI in August: Dwindling hype, unexpected uses, model gardens, and sprawling stacks
What the industry's been talking about for the past month. Plus, a brief conversation with RunwayML's CTO.
I’ll be observing the federal holiday next week, which means there will be no issue on Tuesday. For more information on what holidays I take and how it affects publishing, please check the about page. Wednesday’s issue will also be moved to Thursday to account for a missing reporting day. See y’all Thursday!
We’re closing in on the end of the summer of AI, and as everyone starts to trickle back into the office (Zoom-powered or otherwise) the development and release schedules for a lot of major companies are going to start to pick up considerably.
That more or less kicked off slightly ahead of schedule with Google Cloud Next in August, though summer has hardly been “boring” in the development of artificial intelligence and big data.
Here’s what we’ll be breaking down:
The AI hype cycle drops back and we enter the Era of Practicality. Companies are starting to realize the limitations of current-generation language models and where they can actually use them. And the reality is while coding copilots are pretty good, we’re still looking for that “killer use case” in AI.
Unanticipated powerful uses for generative models. Three projects have come up recently showing how developers are finding some not-necessarily-intuitive positive results from experimentation in AI development.
What does “on premise” look like for AI? With the release of Google’s Vertex AI, we’re starting to see some initial signs of the scaffolding for what “on premise” LLM deployment looks like. And the business models for these APIs are also starting to take shape.
Finally, as we look toward what’s happening this month (and what is on the horizon for the end of the year), the industry seems to be zeroing in on what was a looming problem for years now: the sprawl of the data stack and what’s looming for it for the rest of the year.
As a bonus round, I got to chat with RunwayML CTO Anastasis Germanidis about some of the recent technical developments in generative AI going beyond how it’ll impact the industry with its video-generation tool.
With that, let’s get into it!