The first risk is that they may start doing unwanted tasks (e.g. I ask them to run a marketing campaign for me, and the agents decide to spam everyone, manipulate people, hack into computers, etc)
SAFETY
-
Untested Abilities and Emergent Phenomena in Scaling Large Language Models
By
–
2C. Since we haven't tested all possible abilities, we don't know the full range of abilities that have emerged in large language models.
2D. We're likely to see more emergent phenomena as we continue to scale up models (and implicit argument for more scaling). -
Robot Manufacturers Need Same Liability as Car Makers
By
–
"The manufacturer of a robot should have exactly the same amount of responsibility as the manufacturer of a car." E.g. if a car/robot deployment has a braking malfunction and causes harm, prepare to recall it. And there is a risk of lawsuits.
-
LLM Limitations: Reality vs Hype and Unrealistic Expectations
By
–
1/3 As companies hand tweak LLM based systems to cover published failures those who want to believe in a new super being get more and more sucked in. I use LLMs daily, but that are not atall what many people want them to be, nor what they assume will come just around the corner.
-
Clearview AI scraped billions of images for police facial recognition
By
–
Clearview AI scraped 30 billion images from Facebook and other social media sites and gave them to cops: it puts everyone into a 'perpetual police line-up' US police have used the database nearly a million times, the company's CEO told the BBC. https://
bit.ly/40RUova -
Artificial Intelligence Risks: Bias and Racism
By
–
Risks of artificial intelligence: scientific, cognitive and racist bias #BeAware #ArtificialIntelligence #Risks
-
Artificial Intelligence Risks: Generating False Fake Data
By
–
Artificial Intelligence Risks: The Ability to Generate Incorrect Fake Data #Beware #ArtificialIntelligence #Risks
-
Developer Responsibility for Autonomous AI Agent Actions
By
–
A reminder to folks playing around with BabyAGI and AutoGPT, from a roboticist. Using it to retrieve information is one thing. But if you are actually coding it to DO THINGS and take action on the world, *you* are ultimately responsible for its actions.
-
ChatGPT Bots Enable New AI Experiment Possibilities
By
–
I saw this and realize with all the ChatGPT bots, an "experiment" could already happen https://
x.com/BillyM2k/statu
s/1646375958851768327?t=N6K0ilzz4-V1auvbiGLeZA&s=19
… -
Building Proactive Defenses Against Rogue AI Systems
By
–
I don’t want to stop the development of AI. But I think we need to build proactive defense against rogue AI, just like we build antivirus and firewall SW.