What You May Have Missed #39
Top 5 picks: Hallucinations aren't fixable / Nvidia H100 GPU shortage / Does GPT-4 reason or recite? / Apple has invested billions in generative AI / 10 examples of high-quality text-to-video
Top 5 Picks
Tech experts are starting to doubt that ChatGPT and A.I. ‘hallucinations’ will ever go away: ‘This isn’t fixable’ (Matt O’Brien and The Associated Press on Fortune): “… even [Sam] Altman, as he markets the products for a variety of uses, doesn’t count on the models to be truthful when he’s looking for information for himself. “I probably trust the answers that come out of ChatGPT the least of anybody on Earth,” Altman told the crowd at Bagler’s university, to laughter.”
Nvidia H100 GPUs: Supply and Demand (GPU Utils blog): “In short: Yes, there’s a supply shortage of H100 GPUs. I’m told that for companies seeking 100s or 1000s of H100s, Azure and GCP are effectively out of capacity, and AWS is close to being out.” Andrej Karpathy says “Who’s getting how many H100s and when is top gossip of the valley rn.”
Reasoning or Reciting? Exploring the Capabilities and Limitations of Language Models Through Counterfactual Tasks (Wu et al.): “… while current LMs may possess abstract task-solving skills to a degree, they often also rely on narrow, non-transferable procedures for task-solving.” Chomba Bupe summarizes the study saying it “shows that GPT-4 is more about recitation without attribution than about abstract reasoning.”
Apple has quietly invested billions in generative AI (Cecily Mauran on Mashable): “Apple CEO said its $22.6 billion spending on research and development is driven by investment AI technologies, including generative AI … [which explains] the increase in spending, which has gone up by $3 billion this year.”
10 examples of high-quality AI-generated video (Min Choi):“Creators can now generate cinema quality short films, trailers, and teasers with 100% AI in just a few hours.” Addition (Eric Vyacheslav): Midjourney into Runway’s Gen-2.