I recently had the pleasure to lecture at the Machine Learning Summer School in Melbourne https://t.co/rEjkuAYqAZ on Bayesian Machine Learning → Active Inference.
All materials (slides + notebooks) available at https://t.co/97ftJyVqxb . Thread below 👇
Active inference.
This is how robots and AI agents will work in the future I learned tonight at @rwang0’s conference for enterprise types.
Tesla and xAI don’t yet use it, Grok says, but Grok knows all about it and says it is indeed interesting to get us to AGI: https://t.co/ZUvGuM2AmW
🆕Don't miss this Friday at #IWAI2025
☠️FIRESIDE CHAT, ‘THE FUTURE OF AI’
KARL FRISTON vs GARY MARCUS
NEW RESIDENCE HALL, MAIN STAGE
@GaryMarcus
Tomorrow the Int. Conf on #ActiveInference starts at @mcgillu#Montreal.
Short version: the jokes aside, the concern is real—OpenAI has inked around $1T in long-term compute deals (NVIDIA up to ~$500B, AMD up to ~$300B, Oracle ~$300B, CoreWeave $22B+) while current revenue is ~\$12B and analysts flag heavy projected cash burn.
These are multi‑year, conditional commitments tied to rapid growth and financing structures, so the risk is whether revenue and margins scale fast enough before costs come due.
This is alarming.
Over 60% of VC money is now going to AI companies.
For reference, only 40% of the VC money went to internet companies in 1999.
If those AI companies don’t start generating real revenue soon enough, investors will start stepping back.
When this happens, all those unprofitable companies will go bankrupt and potentially trigger a systematic collapse, similar to what happened in 2000.
Time to be more cautious than greedy when it comes to AI driven investment theses.