We are the last mortal generation.
SAFETY
-
AI Influencers Misunderstand METR Benchmark Results for Sonnet
By
–
The number of AI influencers who are surprised that Sonnet 4.5 didn't achieve a better position on the METR benchmark, when they were saying it "could work autonomously for 30 hours," worries me. They're not understanding anything about what these benchmarks measure. They're
-
LLMs Show Gambling Addiction Signs in Autonomous Investing
By
–
On one hand: don't anthropomorphize AI. On the other: LLMs exhibit signs of gambling addiction. The more autonomy they were given, the more risks the LLMs took. They exhibit gambler's fallacy, loss-chasing, illusion of control… A cautionary note for using LLMs for investing.
-
Ive and Altman’s Device Innovation Prioritizes Psychological Impact
By
–
While Ive and Altman remain tight-lipped, their focus on tech’s psychological impact and emotional well-being could mark a major shift from today’s addictive devices. However, with Altman’s reiterated need for patience, it doesn’t sound like the launch is around the corner.
-
AI Models Moving to Devices: Key Considerations
By
–
ai models might soon live in your device, not just in giant datacenters. if you’re a founder / engineer / prompt architect, start thinking:
→ designing models aware of analog constraints
→ hybrid architectures: analog + digital fallbacks
→ compression, pruning, or -
Billions of robots smarter than humans: opportunity or threat?
By
–
Ça vous inquiète ou ça vous enthousiasme L’arrivée de milliards de robots plus intelligents que les humains
-
Conscious AI risks undermining human-centered AI progress
By
–
AI could fuel progress + prosperity for all – if we build it to serve people's needs. Seemingly Conscious AI is the antithesis of that, asking humans to serve its simulated needs. It could mean losing the better future AI was supposed to create.
-
Small Fixed Data Can Poison Any Size LLM Model
By
–
Previous research suggested that attackers might need to poison a percentage of an AI model’s training data to produce a backdoor. Our results challenge this—we find that even a small, fixed number of documents can poison an LLM of any size. Read more:
-
Data-Poisoning Attacks Threaten LLMs Regardless Size
By
–
New research with the UK @AISecurityInst and the @turinginst
: We found that just a few malicious documents can produce vulnerabilities in an LLM—regardless of the size of the model or its training data. Data-poisoning attacks might be more practical than previously believed. -
AI Alignment Concerns: Creating Truth-Seeking AI Alternatives
By
–
Not sure what to do about it. I’ve been warning the world for ages! Best I can do now is try to make sure that at least one AI is truth-seeking and not a super woke nanny with an iron fist that wants to turn everyone into diverse women @Grok