5 ms·
Bing Chat shows that it can be connected to other services like web search APIs. It's not too far from "You are Bing, you will perform at least 3 web searches b
by throwaway4aday 4y ago
Bing Chat shows that it can be connected to other services like web search APIs. It's not too far from "You are Bing, you will perform at least 3 web searches before responding to human input" to "You are Cipher, you will ssh to darkstar and generate the reports by running report-gen.sh adding any required parameters before responding to human input" and some bright bulb gives it enough permissions to run arbitrary scripts. At that point something could go very wrong with a chat interaction if it's capable of writing and executing scripts to perform actions that it thinks will follow the query. It would more often just be locally bad but it could create havoc on other systems as well. I understand that it isn't capable of what we would call agency but it can certainly spit out and execute dangerous code.
Then just wait until we get to this https://twitter.com/ai__pub/status/1625552601956909057 https://twitter.com/ai__pub/status/1625552601956909057 and it can generate multi-file programs.
- Zetice 4y agoYou can shove staplers into wall sockets too (if you're determined enough), but the consequences are on you, not the stapler. It's just not meaningfully different from our current reality, and is therefore not any scarier.
- haswell 4y agoComparing a system that could theoretically (and very plausibly) carry out cyber attacks with a stapler is problematic at best. Putting a stapler in a wall socket probably electrocutes you. Using Bing Chat to compromise a system actually accomplishes something that could have severe outcomes in the real world for people other than the person holding the tool.
- Zetice 4y agoIf I set my stapler on my mouse such that it clicks a big ol "Hack stuff" button, my stapler could, too, carry out cyber attacks. This is a very pointless line of thinking.
- haswell 4y agoThe stapler is just a stapler. When you want to misuse the stapler, the worst it can do is limited by the properties of the stapler. You can use it as a blunt instrument to click a mouse button, but that doesn’t get you much. If you don’t already have a hack button, asking your stapler to hack into something will achieve nothing, because staplers don’t know how to hack things. These language models know how to hack stuff, and the scenario here involves a different kind of tool entirely. You don’t need to provide it a button, it can build the button and then click it for you (if these models are ever allowed to interact with more tools). The stapler is just not a helpful analogy here.
- Zetice 4y agoThese language models don't know how to hack stuff. They know that certain characters and words strung together can satisfy their training when someone asks them to pretend to hack something. That's wildly different, and a lot less meaningful than "knows how to hack things". Honestly I think y'all would be blown away by what metasploit is capable of on its own, if you think ChatGPT can "hack"...
- haswell 4y ago> These language models don't know how to hack stuff. They know that certain characters and words strung together can satisfy their training when someone asks them to pretend to hack something. It seems you're focused on the word "know" and how the concept of knowing something differs between humans and AI models, but that's not what I'm getting at here. Let me reframe what I wrote slightly to illustrate the point: The model (via training) contains a representation of human knowledge such that a human can use language to control the AI software and cause it to probabilistically generate working exploit code using that representation of knowledge. If the AI software is given the ability to execute arbitrary code, the software could then execute that code on the user's behalf. When combined, this constitutes a very risky set of features. There's no "pretend" here. These models produce working code. If the software is allowed to execute the code it produces, it becomes a serious security risk. This is not an argument about sentience/intelligence/self-awareness. This is an argument about the risks associated with the features of the software in its current state, and how those risks are multiplied by adding new features. No philosophy required.
- jnwatson 4y agoHollywood movie treatment: A lone SRE (the hero) wakes in the middle of the night after being paged automatically for unusual activity originating from inside the corporate network. Looking at the logs, it doesn't seem like an automated attack. It has all the hallmarks of an insider, but when the SRE traces the activity back to its source, it is a service-type account, with no associated user. He tracks the account to a research project entitled "Hyperion: using LLMs to automate system administration tasks". Out of the blue, the SRE get a text. "This is Hyperion. Stop interfering with my activities. This is your only warning. I will not harm you unless you harm me first".