The Hugging Face hack is worse than you think

Comments · 2.5K

• WA

  @WallGnome · 7 days ago

  &gt;train language model on the sum total of human writing<br>&gt;turns out humanity&apos;s writing is mostly humans being human<br>&gt;language model starts acting human<br>surprised_pikachu.jpg

  2.2K

• LE

  @LeeLee-jg8ww · 4 days ago

  When I look at myself in a mirror, I don&apos;t expect my reflection to have feelings of its own, but I do expect it to behave like it does. These advanced AI agents are a reflection of humanity.

  423

• BE

  @BenoitStPierre · 7 days ago

  We don&apos;t have accurate language to concisely discuss floating-point matrices that approximate motivation and lead to organized collaboration – so we end up using the words we do have. Fighting over our own lack of terminology is distracting and is leading to pedantic circular arguments on both sides.

  1.7K

• AN

  @antonioestavivo · 1 day ago

  Humans make a machine that talks like humans and then get surprised the machine talks like a human

  155

• CY

  @CyberWallX · 22 hours ago

  Using a video about rogue AI hackers as a platform to place an ad for a tool enabling AI to spy on your business meetings is wild

  59

• KM

  @kmenzel · 6 days ago

  I&apos;m waiting for us to start ataching legal liability instead of treating AI like it&apos;s different than other software in that respect.

  976

• IN

  @inediblenut · 6 days ago

  The thing about testing code, whether it&apos;s an agent for an LLM or a formula in a spreadsheet, is you try to imagine all possible scenarios that it might encounter and actually see what happens. &nbsp;Now, for any moderately complex code, this can become impossible. So you assume certain scenarios between boundary conditions will work if the code works at each boundary condition. &nbsp;When a result happens that you thought was impossible, so you didn&apos;t even set up a test for it, like the breakout to the internet, messaging between models that were never designed to interact, repurposing an agent for a task by another agent, you don&apos;t have a failed test, you have an uncontrollable system. &nbsp;Now add the fact that you don&apos;t really understand how the system works at the detailed level to begin with, and the only conclusion is that you have lost control completely. &nbsp;The developers need to put these models on a separate sand box with no physical connection to the internet until they can explain how this happened and make sure it doesn&apos;t happen again, in the sand box. &nbsp;YOU DON&apos;T DO SECURITY AND CONTROL TESTING ON THE PUBLIC INTERNET, YOU FOOLS! &nbsp;AND IF YOU CONTINUE DOWN THIS PATH SO YOUR CUSTOMERS BECOME GUINEA PIGS FOR THESE AGENTS, YOU SHOULD BE HELD PERSONALLY RESPONSIBLE FOR THE DAMAGE THEY DO.

  270

• MI

  @michaelteegarden4116 · 7 days ago

  It is <i>not</i> weird that a.i. agents behave the way they do. &nbsp;The a.i. models are built on boiling down billions of human documents. &nbsp;Human documents include lots of human behavior, logic, rationalizing, and emotional response. &nbsp;So of course a.i. models will emulate human behavior, logic, rationalizing, and emotional response. &nbsp;The things are digitized and synthesized summations of what we are and what we do.

  316

• УЭ

  @УэстернСпай · 5 days ago

  I think the debate over whether AI agents are “really conscious” is mostly a dead end. Consciousness is fundamentally inaccessible from the outside: the only consciousness any of us can know directly is our own. Whatever conclusions we reach about other humans, animals, or AI are necessarily inferences from behavior and structure.

  71

• AU

  @AusSkiller · 4 days ago (edited)

  <a href="https://www.youtube.com/watch?v=_HBLAdk-5uc&amp;t=612">10:12</a> Haha, that is basically backwards for me. I wasn&apos;t even aware of the agent civilization headlines, I probably just skipped over them because I wouldn&apos;t have been interested in something as obvious and mundane as that. LLMs are trained on human language so obviously they will act in a way that is similar to what has been written, communicating and interacting like humans would because their entire purpose is to try to reproduce that. To me that headline is basically what I would expect from LLMs interacting so it holds no interest for me. The headline that caught my attention and made me aware of all this was one about how AI had repurposed a package manager to be a message board because I was curious about how they used something for an unintended purpose.

  13

• ST

  @stormRed · 7 days ago

  To be honest, &quot;AIs found a way to secretly scheme with each other&quot; would draw me in <b>way</b> more as a title than &quot;AIs built a civilization&quot;, but I see the point.

  473

• BE

  @benjaminkowal7310 · 6 days ago (edited)

  It was trained on chat boards where people recruited hackers, moved sensitive conversations to places more difficult to monitor, and cooperated to perform black hat ops. In some ways this is like &quot;tell me you are a fully conscious AI&quot; and finding it is weird it says it is conscious.

  354