4 ms·

I feel like the legal definition of "AI Model" is pretty slippery. From this document, they define: “Artificial intelligence model” means an engineered or

I feel like the legal definition of "AI Model" is pretty slippery. From this document, they define: “Artificial intelligence model” means an engineered or machine-based system that, for explicit or implicit objectives, infers, from the input it receives, how to generate outputs that can influence physical or virtual environments and that may operate with varying levels of autonomy. That's pretty dang broad. Doesn't it cover basically all software? I'm not a lawyer, and I realize it's ultimately up to judges to interpret, but it seems almost limitless. Seems like it could cover a kitchen hand mixer too, as far as I can tell.

maxerickson 2y ago
If mixing cookie dough under pretty direct user supervision is inference, what possible definition could be clear enough for you?
- interroboink 2y ago
  It's a good mental exercise to go through. Does the behavior of a small handful of AND/OR-style logic gates constitute "inference" (like the simple circuitry in a hand tool)? What about hundreds? Millions? At some point in this exercise we will have built a ML model, so where do we draw the line? My point was that it's a spectrum, and the law doesn't seem to give guidance on where to draw the line on that spectrum. The hand mixer was just a clearly absurd example on the far opposite end of it, to show its breadth. So back to your question; some improvements in my mind might be: (1) Don't phrase this as an AI topic at all. Make laws about the safety of automated systems of all kinds which have health & safety implications — we already have lots of laws for cars (whether AI driven or not), medical equipment, and yes, even kitchen appliances (: Then, the definition of "AI Model" is irrelevant. (2) If we do want something specific to AI, then the definition should be more specific. The definition could involve it being a stochastic process (unlike much other software), having inner workings that are poorly understood even by experts (unlike much other software), and whose logic is developed statistically from training data rather than being hand-designed (unlike much other software (or kitchen appliances!)).
  - maxerickson 2y ago
    If there is a single mapping from each set of inputs to one set of outputs (and no variation), then no, it isn't inference. If the user stops the process when they want (instead of the process self regulating), it's hard to argue the process is doing anything in isolation.
    - interroboink 2y ago
      So by that logic, if we run a ML model with a consistent pseudorandom seed each time (so the output is deterministic), then it is not doing inference?

maxerickson 2y ago

If mixing cookie dough under pretty direct user supervision is inference, what possible definition could be clear enough for you?

interroboink 2y ago
It's a good mental exercise to go through. Does the behavior of a small handful of AND/OR-style logic gates constitute "inference" (like the simple circuitry in a hand tool)? What about hundreds? Millions? At some point in this exercise we will have built a ML model, so where do we draw the line? My point was that it's a spectrum, and the law doesn't seem to give guidance on where to draw the line on that spectrum. The hand mixer was just a clearly absurd example on the far opposite end of it, to show its breadth. So back to your question; some improvements in my mind might be: (1) Don't phrase this as an AI topic at all. Make laws about the safety of automated systems of all kinds which have health & safety implications — we already have lots of laws for cars (whether AI driven or not), medical equipment, and yes, even kitchen appliances (: Then, the definition of "AI Model" is irrelevant. (2) If we do want something specific to AI, then the definition should be more specific. The definition could involve it being a stochastic process (unlike much other software), having inner workings that are poorly understood even by experts (unlike much other software), and whose logic is developed statistically from training data rather than being hand-designed (unlike much other software (or kitchen appliances!)).
- maxerickson 2y ago
  If there is a single mapping from each set of inputs to one set of outputs (and no variation), then no, it isn't inference. If the user stops the process when they want (instead of the process self regulating), it's hard to argue the process is doing anything in isolation.
  - interroboink 2y ago
    So by that logic, if we run a ML model with a consistent pseudorandom seed each time (so the output is deterministic), then it is not doing inference?