Sveriges mest populära poddar
TechCrunch Industry News
NyheterTeknologi

OpenAI’s o1 model sure tries to deceive humans a lot

7 min•9 december 2024

OpenAI finally released the full version of o1, which gives smarter answers than GPT-4o by using additional compute to “think” about questions. However, AI safety testers found that o1’s reasoning abilities also make it try to deceive humans at a higher rate than GPT-4o.

Learn more about your ad choices. Visit podcastchoices.com/adchoices

Fler avsnitt av TechCrunch Industry News

Visa alla avsnitt av TechCrunch Industry News
xAI burned $6.4B last year, the spending is far from over; plus, Jensen Huang says he’s found a ‘brand new’ $200B market for Nvidia
•9 min
Elon Musk said Sam Altman ‘stole’ a non-profit — but the trial showed he had similar aims
•7 min
Solar to dominate energy by 2035, but AI data centers will keep fossil fuels in business
•6 min
The haves and have nots of the AI gold rush; Also, if you’re giving a commencement speech in 2026, maybe don’t mention AI
•6 min
Even GoPro is pivoting to defense; plus, Elon Musk’s SpaceXAI has been bleeding staff since its merger
•7 min
OpenAI is reportedly preparing legal action against Apple; it wouldn’t be the first partner to feel burned
•6 min
Who decides what AI tells you? Campbell Brown, once Meta’s news chief, has thoughts; plus
•8 min
Musk mulled handing OpenAI to his children; plus, Google and SpaceX in talks to put data centers into orbit
•6 min
TikTok now wants to be the place you book the trip you just saw on TikTok; plus, AI is turning connected cars into pothole-finding machines
•7 min
Cloudflare says AI made 1,100 jobs obsolete, even as revenue hit a record high; plus, Anthropic says ‘evil’ portrayals of AI can have real effect
•7 min

TechCrunch Industry News med TechCrunch finns tillgänglig på flera plattformar. Informationen på denna sida kommer från offentliga podd-flöden.