thisoneworks

Why would you trust a models opinion about its own possibly emergent ability?

I would disagree with this. There have been several instances where the effective use of this First-Person language has had measurable gains in success rates across tasks. The effective usage of this First-Person language for communication and steering between a model/agent and a human is entirely different than whether the model itself can "think" - that does not matter for the former to happen.

It's crazy how far i had to scroll to find someone saying exactly what i was thinking - how come AI sloppy writing like this is showing up on Hn front page?

I'm sure there are some good advice in there but lots of slop as well

Chill my dude. This is just a sane default which will catch normies copy pasting stuff from claude and chatgpt. It's good enough.

"Give everybody a gun! So they can defend themselves from the bad guys! Don't let the government take away your guns!!"

I mean what are they doing right now at Google? Optimizing data centres? Pretty lame compared to this. Even if they completely fail, i'm sure there'll be good lessons.

Imo it definitely has to do with politicians and governments trying to appear strong on the topic of protecting kids from the harms of social media. I also believe a lot of it is well intentioned, albeit poorly executed

It'll be funny when we have Robots, "The user's facial expression looks to be consenting, I'll take that as an encouraging yes"

You can frame the "unpaid voluntary labor" as "creative work" and it would start making a whole lot of sense. "Creative work thrives despite being unpaid in capitalist society."

Hard disagree. You can't just one day wake up and double your energy infrastructure.. China is way ahead.

Regardless of the usefulness of llms, if you don't work at anthropic, how gullible are you to believe that claim at face value?

Big part of my annoyance is the term "AI" itself which you can say to mean anything, everything and nothing. It's something that's used to oversell/hype it to grab money.. which is fine. But if engineers like us can calm the crazy rhetoric down to "llms" or "text completion" "Image-gen api" that's already a leg up in thinking clearly. Like the question "is ai going to put people out of jobs" -> "is this new crazy good text completion model going to put people out of jobs" already gets us out of the weeds somewhat.

Hah financialization strikes again. Try explaining this to a person from a third world country, they would say "what are you talking about". Also they would have better health care than your average American.

"gemini for video games" - here we go again with the AI does the interesting stuff for you rather than the boring stuff

Since when did naming a country for their military action signify the opinion or inclination of the majority of civic population? When newspapers report on "country A did X" it almost always means their government did X. So I'm not sure what point you're trying to make

I honestly don't think this is all that big. What we are seeing has been possible for more than 6 months now(?) with gpt4 and elevenlabs, its just put together in a nice little demo website and with what seems like a multi-modal model(?) trained on nytimes the daily episodes lol. And no i don't think this will gain all that much traction. We will keep valuing authentic human interaction more and more.

Yeah that wasn't obvious what they were trying to show. Demis said feature films will be released in a while

Add Slack to that list

That's not 8bit..

I don't see anything in https://docs.github.com/en/search-github/github-code-search/... :/

Instead of relying on one big model (and all it's flaws), aren't you better off having separate smaller models? Better for auditing

At that point I'm pretty sure they'll already have fine tuned a "senior software engineer agent". This may sound ridiculous and yeah we probably won't get rid of the entirety of SWE ladder, but my point is we are entering the realm of science fiction now, our assumptions about labor are about to fall off a cliff. Productivity will increase through the roof, corps will make more cash, workers will suffer

You hit the nail on the head. GPT/copilot in some sense is democratization of specialized knowledge. Now on one end you'll have product engineers/managers prompt gpt to write boilerplate code of all kind, on the other end you'll have senior engineers reviewing, optimizing writing specialized code. Where do junior engineers fit? Can you be a full stack engineer? I believe it'll squeeze out all the could-be devs out of the field forever. Gender diversity/DEI? Forget it. Upward mobility? Will become much harder, you'll basically need to have a specialization to even be considered

There might be government regulation on AI pretty soon.. it's not crazy to think GPUs and GPU tech would be treated as defense equipment some day

You're better off placing your bet on Russian and Chinese hackers, crypto scammers than a Joe Shmoe. But read https://aisnakeoil.substack.com/p/the-llama-is-out-of-the-ba... - there's no noticeable rise in misinformation

Oh I'm aware of it. I do not think it holds any merit right now when we're talking about coding assistants.

With great power comes great responsibility? Today there's nothing stopping grandmas from driving, so whatever could go wrong is already going wrong

Who are "we"?

You want guaranteed specific behavior from a software system (that's what SLAs and contracts are for) and easily reason about it so you can hire a college grad to tweak it. And that's not even talking about datasets, you can only train something accurate if you have enough good data on it? I can only see software 2.0 being a better autocorrect for internet scale usecases (in the short term atleast, until the next breakthrough)

This also reminds me of a quote from the Book I'm currently reading (Practical Wisdom by Barry Schwarz)

“Most of us think about empathy as a “feeling” or an “emotion.” It is. To be empathetic is to be able to feel what the other person is feeling. But empathy is more than just a feeling. In order to be able to feel what another person is feeling, you need to be able to see the world as that other person sees it. This ability to take the perspective of another demands perception and imagination. Empathy thus reflects the integration of thinking and feeling.”

"Mind reading" is another way to put it (https://yosefk.com/blog/people-can-read-their-managers-mind....) - this practical wisdom + mind reading is basically the salient human feature that NNs would never be able to replace so you would always have humans in the system.