The field of AI safety is in dire need of reliable data .
The UK AI Safety Institute is poised to conduct studies that will hopefully bring hard data to a field that is currently rife with wild speculations and methodologically dubious studies.
SAFETY
-
UK AI Safety Institute Launches Reliable Data Research Initiative
By
–
-
AI Safety, LinkedIn Growth, ChatGPT Traffic, Instagram AI Friend
By
–
Les sources de l'édition du jour du #FlashTweet https://
reuters.com/technology/lin
kedin-hits-1-billion-members-adds-ai-features-job-seekers-2023-11-01/
… @reuters https://
techcrunch.com/2023/11/01/pol
iticians-commit-to-collaborate-to-tackle-ai-safety-us-launches-safety-institute/?utm_source=pocket_saves
… @techcrunch https://
maddyness.com/2023/11/01/doo
itch-reinvente-la-logistique-du-secteur-du-btp-grace-a-son-service-de-livraison-express/?utm_source=pocket_saves
… @byMaddyness https://
digitalinformationworld.com/2023/11/chatgp
ts-web-traffic-is-back-on-track.html?m=1
… @DigitaliWorld https://
techcrunch.com/2023/11/01/ins
tagram-spotted-developing-a-customizable-ai-friend/
… @TechCrunch -
AI Safety Summit: GPAI Members and International Collaboration on Frontier AI
By
–
Half the attendees at the #AIsafetySummit were #GPAI members and the rest are countries that we are establishing relations for collaboration because our mission is to achieve an AI that has assurance and is truly a tool for progress of all, the motto of our next President chair Department for Science, Innovation and Technology (@SciTechgovuk) 28 countries & the EU have signed The Bletchley Declaration at the #AISafetySummit agreeing to: ⚠️ identify the key opportunities & risks of AI 🌍 build a global understanding of Frontier AI risks 🔬 collaborate on AI scientific research Find out more: gov.uk/government/publicatio… — https://nitter.net/SciTechgovuk/status/1719728536460710027#m
→ View original post on X — @inma_martinez, 2023-11-02 06:58 UTC
-
Llama Open Source Release Ethics Phil Zimmerman Comparison
By
–
I appreciate the compliment, but I'm not sure I deserve it.
Unlike Phil Zimmerman, who wrote the code for PGP and risked imprisonment by releasing it, I didn't write Llama, and I didn't risk imprisonment! -
R&D Risk Assessment and Concentration in AI Development
By
–
I already believe R&D can be high risk, at least under some definition of that term. What makes you think I believe otherwise? (I've repeatedly pointed out that it's likely much *more* high risk if this is highly concentrated.)
-
Llama 2 Safety: Evaluating Meta’s Open-Source Model Deployment
By
–
Sometimes instead of hypotheticals, it’s better to take concrete practical examples. Let’s try with Llama 2 by @Meta (
https://
huggingface.co/meta-llama). What do you think would have been the safest scenario: -
Existential AI Threats and Normalization Concerns Debate
By
–
Attacking my concerns about the existential threat by saying that I am normalizing or cheering on genocides is an interesting approach.
-
Expert Input Essential for Nuclear Safety Policy Discussions
By
–
You're putting words in my mouth, and totally misunderstanding the point. I did not say I think we never get there. I simply said that experts on nuclear safety should be considered when discussing nuclear safety.
-
Nuclear Experts Should Be Heard in AI Risk Discussions
By
–
I agree – I certainly don't want him to stop speaking. But I do think nuclear experts should also be listened to when considering nuclear analogies. There are many other experts, in fact, with relevant expertise.
-
No Great Precedents Exist for AI’s Transformative Impact
By
–
There aren't any great ones — in the sense there isn't some existing thing we could borrow and copy over all our understanding and experience and expect it to "just work". Somewhat similar things include: nanotech; crispr; LHC; recombinant DNA; cryptography.