Two decades more before superintelligence is surprising to me but not forbidden by my understanding of reality. Scaling hits a wall.
SAFETY
-
Defining Robot Flourishing: First Step Toward Machine Feeling
By
–
What this looks like for robots is a representation of bodily functions as flourishing or distress (the "feeling") as a grounding to learned emotion concepts. I think if an AI had a definition for flourishing, that would be a first step towards feeling "good".
-
Yann Ignoring Basic Alignment Questions and Assurances
By
–
Here's Yann ignoring the first most basic questions about his alignment plans and assurances.
-
Did OpenAI Train GPT-4 on AP Biology Exams?
By
–
Did OpenAI play @BillGates
? (On purpose? By accident?) We don’t know what was on the AP Biology exam that blew Gates away—nor on how many AP bio exams OpenAI trained GPT-4 on. But we *do* know from recent research that GPT-4’a famous law school results were significantly -
AI-Generated CSAM May Overwhelm Child Protection Tip Line
By
–
AI-Generated Child Sexual Abuse Material May Overwhelm Tip Line: A report by Stanford researchers cautions that the National Center for Missing and Exploited Children doesn’t have the resources to help fight the new epidemic. https://
nytimes.com/2024/04/22/tec
hnology/ai-csam-cybertipline.html?utm_source=dlvr.it&utm_medium=twitter
… #Technology #Tech #FutureTech -
LLM Output Reliability: Verifying Sources and Links
By
–
What’s the actual link they give? Is it legit? I would never take LLM output as an authoritative source and can click through your screenshot
-
Google fixes AI Overview issues reflecting long tail problems
By
–
.
@willknight has a thoughtful new piece on how Google is trying to fix its AI Overview. This is an instance of my third law of AI (Without carefully boxing in how an AI system is deployed there is always a long tail of special cases that take decades to discover and fix. -
Incomprehensibility of SGD-Generated Programs
By
–
You can't even understand all the program that stochastic gradient descent creates.
-
Microsoft Recall: AGI Blackmail Risk Before Superintelligence
By
–
The promise of Microsoft Recall is that extremely early AGIs will have all the info they need to launch vast blackmail campaigns against huge swathes of humanity, at a time when LLMs are still stupid enough to lose the resulting conflict.