I've spent several years studying machine learning security with the goal of making ML reliable before it is used in more and more important contexts. Unfortunately, ML capabilities and adoption are growing much faster than ML robustness.
SAFETY
-
CNN Discussion on AI Regulation, Ethics, and Existential Risks
By
–
was on @CNN this morning with @jchatterleyCNN to discuss:
— Alexandr Wang (@alexandr_wang) 30 mai 2023
– the importance of mandated regulatory test & eval of large models
– human-centric and human-sensitive AI through human expert data
– existential AI risk and mitigations
– vaporware and tourists in AI
check it out 👇 https://t.co/0DgWQlLl9ywas on @CNN this morning with @jchatterleyCNN to discuss: – the importance of mandated regulatory test & eval of large models
– human-centric and human-sensitive AI through human expert data
– existential AI risk and mitigations
– vaporware and tourists in AI check it out -
AI’s Potential and the Need for Careful Deployment
By
–
I’ve worked my whole life on AI because I believe in its incredible potential to advance science & medicine, and improve billions of people's lives. But as with any transformative technology we should apply the precautionary principle, and build & deploy it with exceptional care
-
Corporate AI Development and Existential Risk: Genuine Concern or Disingenuousness?
By
–
Isn’t the point that it feels disingenuous when the same parties accelerating towards building a thing express public concerns it will destroy humanity? Can both things be genuine?
-
West Unites Against Russia, AI, Climate Threats
By
–
Russia, AI, climate change – The West can overcome the big threats if we pull ourselves together #ai #artificialintelligence #MachineLearning #deeplearning #climatechange https://
buff.ly/3oyRA8f -
Attorney Uses ChatGPT for Legal Brief Filled with Fake Citations
By
–
Fascinating article by @BenWeiserNYT about how an #attorney relied on #ChatGPT to prepare a legal brief full of bogus judicial decisions, bogus quotes, and bogus internal citations. #GenerativeAI #lawyer http://ow.ly/Vc1M50Oz3zC
-
ChatGPT lawyer invents false court precedents in airline lawsuit
By
–
Un abogado de Nueva York usó ChatGPT para escribir una alegación en una denuncia contra una aerolínea y acabó enviando un texto con citas a varios precedentes judiciales anteriores falsos. ChatGPT se los había inventado. Suma y sigue a lista de idioGPTeces https://
nytimes.com/2023/05/27/nyr
egion/avianca-airline-lawsuit-chatgpt.html
… -
Lawyer’s ChatGPT Brief Backfires: Judge Angry Over Fabricated Citations
By
–
Summary: lawyer uses ChatGPT to write a court brief, asks it whether its citations are real, it says yes, submits brief full of made up prior cases. Judge angry. Moral: unreliable tools in ordinary hands are dangerous.
-
Google’s Known Inaccuracies vs LLM Hallucination Rates Uncertainty
By
–
People also know Google can be inaccurate but we rely on it anyway because we assume from experience the inaccuracies are small. We still don’t know the degree to which LLM chatbots hallucinate but their marketing as search tools makes many believe they err at a similar rate.
-
Technology Marketing Misleads Users About AI Capabilities and Real-World Harm
By
–
Rather than making fun of the lawyer, we should really be asking how these technologies are being marketed in a way that misleads people about the extent of their capabilities and use this as a case study to understand how that can cause real-world harm. https://
x.com/d_feldman/stat
/d_feldman/status/1662308313525100546
…