I think AI now officially qualifies as a “sufficiently advanced technology”
@erikbryn
-
Census Data Analysis Reveals Patterns in Robot Usage
By
–
Yes, I agree. We're seeing a similar pattern in the census data we're analyzing on the use of robots, e.g. https://
aeaweb.org/articles?id=10
.1257/pandp.20231038
… -
Extinction risk statement signatories and Yann LeCun’s critical stance
By
–
Hadfield and Song both signed the extinction risk statement. Yann LeCun is probably the single most prominent critic of it. And yes, his views like almost everyone’s are pretty nuanced. And to be clear, there are many other prominent people on all sides of this discussion.
-
Critique of cherry-picking names and call to transcend divisions
By
–
1. Cherry-picking names is misleading and unfair to folks like Mira Murati, Gillian Hadfield, Pattie Maes, Dawn Song, Victoria Krakovna & Yann LeCun 2. More importantly, we'd all be better off if smart people like you spent less time trying to tribalize & name-call (e.g. "panic")
-
Preparing for Significant Economic Disruption and Skills Demand Changes
By
–
A more accurate version of the meme would put most of those folks on the right next to me We agree that mass unemployment is unlikely anytime soon, but that we should be ready for significant disruption as demand for skills and tasks change.
-
AI Capabilities Beyond Simple Statistics and Human Understanding
By
–
It's as silly to say "AI is just predictive stats" as it would be to say "neurons are just electrochemical pulses". AI clearly can solve all sorts of very hard problems, which is amazing and awesome. AI capabilities are also, in many ways, increasingly beyond our comprehension.
-
AI Problem-Solving Abilities Beyond Human Capabilities and Unpredictability
By
–
I'm not sure what you mean about 'waking up'. I do believe AI systems will have problem-solving abilities far beyond humans on many, perhaps most, dimensions and that in many cases, AI systems will do things that we didn't anticipate.
-
AI as Primary Romantic Partner or Trusted Friend by 2030
By
–
In the year 2030, what percentage of people in the United States will consider an AI system (e.g. a personalized LLM) to be their primary romantic relationship or most trusted friend?
-
Synthetic Data Effectiveness: Practical Applications and Limitations
By
–
OK that’s a much more defensible claim. It will be interesting to see whether synthetic data works in practice, and if so, where it is most and least effective. But we can’t rule out its potential a priori.
-
From Implicit to Explicit: AI and Mathematical Discovery
By
–
Alphazero, trained on synthetic data, was able to discover novel knowledge that enabled it to defeat real world human Go players. Likewise, number theorists can create new knowledge, though perhaps it was “implicit” all along. Going from implicit to explicit is valuable.