[-] Aidan@lemm.ee 12 points 11 months ago

In one of my 300 level poli sci classes, literally one of the first things the professor said is that in politics, everyone running for office is a power-hungry narcissist. It’s only a slight exaggeration.

That type of person is at every level of politics. I’d wager that if you could get data on the real motivations of every person who has ever run for office, you’d probably see the same amount of those people at every level, from school board to president.

[-] Aidan@lemm.ee 24 points 11 months ago

I like the idea of never referring to it again

[-] Aidan@lemm.ee 25 points 11 months ago* (last edited 11 months ago)

I don't agree that ChatGPT has gotten dumber, but I do think I’ve noticed small differences in how it’s engineered.

I’ve experimented with writing apps that use the OpenAI api to use the GPT model, and this is the biggest non-obvious problem you have to deal with that can cause it to seem significantly smarter or dumber.

The version of GPT 3.5 and 4 used in ChatGPT can only “remember” 4096 tokens at once. That’s a total of its output, the user’s input, and “system messages,” which are messages the software sends to give GPT the necessary context to understand. The standard one is “You are ChatGPT, a large language model developed by OpenAI. Knowledge Cutoff: 2021-09. Current date: YYYY-MM-DD.” It receives an even longer one on the iOS app. If you enable the new Custom Instructions feature, those also take up the token limit.

It needs token space to remember your conversation, or else it gets a goldfish memory problem. But if you program it to waste too much token space remembering stuff you told it before, then it has fewer tokens to dedicate to generating each new response, so they have to be shorter, less detailed, and it can’t spend as much energy making sure they’re logically correct.

The model itself is definitely getting smarter as time goes on, but I think we’ve seen them experiment with different ways of engineering around the token limits when employing GPT in ChatGPT. That’s the difference people are noticing.

[-] Aidan@lemm.ee 55 points 11 months ago

This is one of the problems with using country TLDs. They look cute, but when you buy it, you may not realize who controls it. Lemm.ee is similarly in a precarious position.

I really wish we could all agree to stop using country TLDs for this

[-] Aidan@lemm.ee 14 points 11 months ago

Someone close to me is a HS teacher. During covid, the schools changed their policy from “no phones in class ever” to “you can have your phone in class but you’d better only use it to help with classwork or in an emergency.”

They’ve been trying to reverse the policy back to how it was, but it’s hard to get all the kids to believe that they can’t do this anymore. They don’t take the threat of punishment seriously because everyone is doing it now.

Even if you manage to deal with the phone issue, the school gives kids chromebooks now to do their work on. The student wifi network seemingly has no restrictions, since the teachers sometimes need to have them watch something on YouTube or Netflix.

So kids, during class, watch Netflix on their Chromebook instead of paying attention.

[-] Aidan@lemm.ee 95 points 11 months ago

They sell your data and don’t feel bad. Why should you feel bad about selling your data?

[-] Aidan@lemm.ee 11 points 1 year ago

This interview makes me wonder what’s going to happen after he’s gone. You could say that he’s set up whoever succeeds him for a tough act to follow. But no one necessarily has to succeed him in any way.

[-] Aidan@lemm.ee 10 points 1 year ago

Well you managed to get the username aidan@lemmy.world before me, so that’s something

[-] Aidan@lemm.ee 17 points 1 year ago

The iCloud support app? I’ll say it if you won’t. Apple needs to be shamed into doing something about that

[-] Aidan@lemm.ee 17 points 1 year ago

Working at the morgue must have been tough

[-] Aidan@lemm.ee 18 points 1 year ago

I think we’ll see a variety of servers with different funding models, similar to how radio and tv stations in the us can have a variety of funding models. NPR has a network of member stations that all carry their content (if the stations want, or they can get content from another station, or they can make it themselves).

Threads is an example of a federated service with a corporate funding model. I definitely think it’ll survive since they have as much money as Facebook wants to sink into it.

But we’ll probably also see servers that run on donations by a dedicated community.

If Threads is the NBC/CBS/ABC of the federated landscape, then those small servers will be like public radio stations, which operate on donations and the occasional government grant.

I think there are people who would chip in a little bit to fund a non-commercial server just the same as there are people who chip in money to NPR.

[-] Aidan@lemm.ee 45 points 1 year ago

What part of threads could they even argue was stolen from Twitter? It’s an open source protocol plus Instagram logins. You think Facebook needed to hire people to tell them about how to post text on the internet?

2
view more: next ›

Aidan

joined 1 year ago