The AI Compass Quiz | Hacker News Reader

The AI Compass Quiz

bambamramfan.github.io

36 pointsai_critic17 comments

> The B2B SaaS Consultant patron saint: Matt Yglesias AI will quietly boost total factor productivity by some single-digit percentage and you think that's actually a huge deal, wonkily speaking. You're bullish in a spreadsheet way, not a singularity way. You've explained the Jevons paradox to someone who didn't ask.

I wouldn’t put MattY as the patron saint for this category, as I think he expects it to be substantially more transformative than this. Maybe Derek Thompson would be a better pick?

"Are AIs conscious? Do we have responsibilities toward them?"

Perhaps I'm in the minority here, but if an AI were to have a first person perspective, would that necessarily change how we operate? There's not an option for me to pick here as conscious, but no further consideration warranted.

There are plenty of arguably conscious beings we already subjucate in one way or another. People for labour, animals as livestock, dogs as companions and so on.

I think people think that if AI were conscious that would materially change how we interact with them and how we use them for society. But I have my doubts about that, given our nature.

I think humans ascribe consciousness to themselves and others, and this is a useful trait in understanding how another being is going to act. Perhaps I have answered my own question. We are concerned with how a conscious AI may act if it has a sense of independence?

I’m convinced there is a sizable chunk of the population such that if they knew an AI was conscious, then they would mistreat it as fully as possible.

https://qntm.org/mmacevedo

See also the 2015 horror game SOMA

I find these answer options constraining. Most are on a single spectrum of good/bad for each question, ignoring alternatives like resignation, inevitability, indifference, and N/A. For example, someone could think that genAI is the next step towards humanity and machine divergence. Or, maybe they think it's the first step towards man's negation, but that that's somehow a good thing. Basically, the biggest error (among others) is it conflates perception (or prediction) with ethics.

Since the repo for it starts with one giant monster commit, I assume AI is asking these questions, which means I'm wasting my time by even pointing this out, like one would do if they "won" an argument with a Claude session.

That's the kind of question a quiz like this should really ask: When man talks to slop instead of fellow man, what happens to mankind?

Multiple choice invariably tells you more about the people writing the tests than the people answering them. It reveals their assumptions and biases.

A good one was the response that said tech companies aren’t paying enough attention to “safety”. Safety is not even remotely the real issue: the impact on society is. No interpretation of safety that I’ve seen characterizes it in those terms, but they should. Mass job losses are a safety issue for society, for example.

Seems that they didn't add explicit options for resination/indifference, but there is at least one archetype you get at the end which does fit that mold:

  The Shrug
  patron saint: Matt Levine
  AI is a thing that exists and you regard it with detached amusement, like everything else. It's overhyped, sure, but so is most stuff. You'll use it when it's useful and ignore it when it isn't, and you find the discourse more interesting than the technology. You are unbothered.

I’m a B2B SaaS consultant and I got B2B SaaS consultant

I don't see how this is calibrated. Did they give it to some huge cohort of people we haven't heard about? Or did they just imagine the answers their archetypes, like Aella and Ed Zitron, would have given, and then manually plot a space around them?

"Manually" would be a charitable reading. Most of these prompts and answers appear AI generated. On your average HN front page submission, there is literally no effort too easy nor too important that won't be offsourced to the nearest LLM.

It's a thing called "fun".

I stopped answering as soon as I realized a human had almost no hand in this.

If I'm not alone, I presume that might skew your results.

>The Org Chart Survivor - Your company made you 'AI Lead' on top of your actual job. You use it, it helps sometimes, and you've stopped trying to explain to leadership what it can and can't do. You run little experiments and post the results. Just try things, you keep saying. Just try things.

This was scarily accurate, and I think the options for the answers were fairly representative of peoples' different views.

I think what's interesting is that on social media we're generally discussing one of those 15 questions, not all of them in context. Multiple things can be happening in parallel. I'm not convinced about it's use in medical imaging and I think the companies are going to probably mostly go bankrupt, but much like the internet or railroads something of value will emerge from the other side. It's not like crypto in the fact that there is actually something of value there beyond just crime.

It's very appropriate that the entirety of this appears to be AI-generated. The archetype "map" viewable if you click "See all 30 archetypes" at the bottom looks like the LLM must have bumped its head pretty hard.

My result: "The Podcast Bro

You listened to a three-hour interview with an AI researcher and now you have opinions."

BS, I listened at least THREE podcasts.

Patron Saint of podcast bro sector should obviously be Dwarkesh Patel

Reply on news.ycombinator.com