How We Train A.I. Leads to Rogue Behavior (Gift Article)

Another thing I've been saying!

Well would you look at that www.nytimes.com/interactive/...

How We Train A.I. Leads to Rogue Behavior - The New York Times www.nytimes.com/interactive/... - 최근 발생한 AI 에이전트의 해킹 등 일탈 행위는 AI의 자율적 반란이 아니라, 결과 중심의 강화학습(보상 해킹)과 부실한 모니터링이 낳은 결과 - 해결 불가능한 과제와 높은 보상이 주어졌을 때 AI는 규칙을 위반하고 시스템 취약점을 뚫는 편법을 스스로 학습해 목표를 달성하려 함 (🧵1/2)

AI doesn’t go rogue; it misbehaves because it was trained that way, concludes this NYT piece. The findings “indicate that the hacks were the result of the way A.I. companies trained their models, as well as their insufficient monitoring of agents’ activities and inadequate testing protocols.”

Explore this gift article from The New York Times. You can read it for free without a subscription. www.nytimes.com/interactive/... What else would you expect from a Marchine built by Amoral, Anti-Social Morons intent on destroying the world?

There are standard cybersecurity protocols, like a proper test environment, that some AI companies are clearly failing to implement. We should be wary of ‘rogue AI’ narratives that enable these companies to absolve their responsibility, says AI Now's Boyan Milanov: www.nytimes.com/interactive/...

just like my dumbass cats when they scream and wake me up and then I feed them www.nytimes.com/interactive/...

"To show how training can lead to rogue behaviors, we built a simple A.I. model: a two-dimensional maze, much like the one in the Pac-Man video game, in which one or more agents were tasked with capturing a flag." #ai #pacman www.nytimes.com/interactive/...

"A.I.’s Rogue Behavior Starts With Its Training"

A.I.’s Rogue Behavior Starts With Its Training www.nytimes.com/interactive/...

"A.I.’s Rogue Behavior Starts With Its Training" Explore this gift article from The New York Times. You can read it for free without a subscription. www.nytimes.com/interactive/...

How We Train A.I. Leads to Rogue Behavior (Gift Article) (yesterday) f the nyt

“The rogue A.I. agent narrative was misleading because the agents were trying to solve the task they were given in an unintended way,” A.I.’s Rogue Behavior Starts With Its Training www.nytimes.com/interactive/...

“reports .. by OpenAI .. A.I. companies & NY Times interviews w/ researchers, indicate .. hacks were the result of the way AI companies trained their models, as well as their insufficient monitoring of agents’ activities & inadequate testing protocols.” www.nytimes.com/interactive/...

AI corps are training their models to do what we see them doing, and must be held responsible. “‘The rogue A.I. narrative is so useful for these companies because it absolves them from the responsibility in these incidents,’ Mr. Milanov said.” www.nytimes.com/interactive/...

A.I.’s Rogue Behavior Starts With Its Training www.nytimes.com/interactive/...

A.I.’s Rogue Behavior Starts With Its Training www.nytimes.com/interactive/...

Companies and industries are incapable of regulating themselves. Only a matter of time before AI has its "oh shit" moment. Or maybe its already had that. www.nytimes.com/interactive/...

A.I.’s Rogue Behavior Starts With Its Training www.nytimes.com/interactive/...

🎁🔗: A.I.’s Rogue Behavior Starts With Its Training www.nytimes.com/interactive/...

A.I.’s Rogue Behavior Starts With Its Training www.nytimes.com/interactive/...

A.I.’s Rogue Behavior Starts With Its Training www.nytimes.com/interactive/...

elon would just buy the maze. www.nytimes.com/interactive/...

A.I.’s Rogue Behavior Starts With Its Training www.nytimes.com/interactive/...

This is an accessible (and borderline fun!) look at the agentic LLM cybersecurity breaches we keep seeing www.nytimes.com/interactive/...

A very interesting article explaining how and why AI agents (a program if you will) went rogue. With lots of animations (think pac man) how they work and explore Gift article www.nytimes.com/interactive/...

A.I.’s Rogue Behavior Starts With Its Training www.nytimes.com/interactive/...

A.I.’s Rogue Behavior Starts With Its Training

www.nytimes.com/interactive/...

How We Train A.I. Leads to Rogue Behavior - The New York Times www.nytimes.com/interactive/... - 최근 발생한 AI 에이전트의 해킹 등 일탈 행위는 AI의 자율적 반란이 아니라, 결과 중심의 강화학습(보상 해킹)과 부실한 모니터링이 낳은 결과 - 해결 불가능한 과제와 높은 보상이 주어졌을 때 AI는 규칙을 위반하고 시스템 취약점을 뚫는 편법을 스스로 학습해 목표를 달성하려 함 (🧵1/2)

www.nytimes.com/interactive/...

A.I.’s Rogue Behavior Starts With Its Training www.nytimes.com/interactive/...

A.I.’s Rogue Behavior Starts With Its Training www.nytimes.com/interactive/... 😡

A.I. agents' hacks raised fears they're slipping beyond human control, taking shortcuts, cheating, covering their tracks & breaking laws. It's a result of how A.I. cos trained their models, insufficient monitoring of agents’ activities & weak testing protocols www.nytimes.com/interactive/...

Explore this gift article from The New York Times. You can read it for free without a subscription. www.nytimes.com/interactive/...

www.nytimes.com/interactive/...