Member-only story
The State of Generative AI, 2024
A nuanced analysis and a glimpse of the future
8 min readApr 3, 2024

This article is a selection from The Algorithmic Bridge, an educational project to bridge the gap between AI and people.
If you watch the news, it’s easy to miss the forest for the trees. The evidence is all scattered so let’s gather it in one place:
- Growth, revenue, and margins are underwhelming. Visits to AI sites have stalled. Sequoia, a VC firm, estimates that in 2023 companies spent $50B on Nvidia hardware but only brought in $3B in revenue. AI startup valuations have been much higher than they should be. The low gross margins raise questions about profits and cloud providers are tamping down expectations.
- High-profile startups are starting to fail. InflectionAI, a well-funded private model-building company, is being dismantled; Microsoft is picking up the pieces, including ex-CEO Mustafa Suleyman. StabilityAI’s future is unstable, to put it lightly, after founder Emad Mostaque’s downfall.
- Enterprises have security doubts and deployment doubts. OpenAI’s GPT store is an utter failure. After an initial spark of interest, people are getting bored. NYT journalist Ezra Klein says he can’t figure out how to use the tech in his day-to-day job. Economist Tyler Cowen says the use of the tools among his fellow academics “has plateaued.” Experts consider AI tools a “fun distraction” but not useful or productivity-enhancing (and when they are, it doesn’t always go well).
- Finally, most people don’t really care. The general public knows about ChatGPT but nothing else. Among those who do, the vast majority use the obsolete GPT-3.5, unwilling to swap to better successors, like GPT-4, Gemini, and Claude 3 for a few bucks. Among those who do, it’s the unscrupulous who are sadly getting the most value — to the detriment of people, the internet, and our culture.
Does this mean generative AI is failing? Perhaps, but not necessarily.
Let’s make a fair case:
- The mainstream’s interest in AI has decreased but AI bubbles still share daily updates read by hundreds of thousands (including this newsletter, which keeps growing at a quick pace). The overhype has slowed down as well, with big names like Google…