I know people often define intelligence as the ability to achieve goals because the AI field defines it somewhat like that, but when communicating with outsiders, I suspect you shouldn’t start the conversation by trying to change how they use the handles to understand the world.
For the average person, intelligence is primarily understood as the capacity to understand and learn new information easily.
We can link what I think are the two most common usages of the term:
As you can see, the everyday usage and the technical definition used within AI safety are substantially far apart.
Technically, not even this definition of the AI Field means the agent would be better than any others at accomplishing goals, because that would depend on its initial conditions too. And that coincides with the way the term is used: an AI model is always compared with others in the same conditions through different environments to define it as "more intelligent".
In conclusion, is really useful to achieve goals, but it is far from being "the ability to achieve goals".
If we want to help their maps of the world, I think there are better alternatives than messing with the handles they put on concepts. It must feel a bit fishy that someone who is trying to make a world-shattering claim is basing all of it on extrapolating from a term that they are using differently than you, Spock is your idea of smart, and you haven't ever thought about the achieving-general-goals property in much detail before.
But I've heard multiple people doing this in the AI Safety communications space.
Instead of insisting that "intelligence IS the ability to achieve goals," we can use clearer formulations that convey the intended message without semantic friction:
One might argue that insisting on “intelligence is goal achievement” creates a simpler shorthand for associating advanced AI with danger, but I doubt it's better.
This text was written with the assistance of Gemini and Grammarly.
Edit: I added the text and formula that consider the Initial empowerment as a factor.
I agree with this post, although I think your suggested alternative #1 runs into the same problem - it just substitutes "smarts" for "intelligence". I think both your #2 and #3 alternatives are better.
I think that the fact that "smart" is being applied to a "digital species" changes things because "species" implies agency & goal-directedness, and "digital" implies it will be able to access useful information and become knowledgeable.
But I can understand that maybe other species don't seem goal-directed, so maybe it feels less threatening.
I see what you mean, but in my experience the people who push back hardest against the "intelligence" label seem to also be the ones who reject framing of AI as a new "species". I'm thinking of the people who still argue AI is "just" a next token predictor, and that we shouldn't "anthropomorphise" them, etc.
So I think your #2 and #3 alternatives would be better at reducing semantic friction with that group.
Yeah, that's true