Author | Lightnews

Cyborg Tibe @cybrogtribe.bsky.social · 5h

Dammit. Now I need to buy new shirts

Reposted by Cyborg Tibe

Menotti Minutillo @menotti.bsky.social · 9h

ladies and gentlemen...we got him

140 2.8K 13K

Reposted by Cyborg Tibe

verdverm @verdverm.com · 1d

Run independent @atproto.com networks, for testing and beyond, in docker compose or kubernetes, public or private

Pushed significant updates today #atdev #atprotodev

github.com/blebbit/test...

GitHub - blebbit/testnet: Run independent ATProtocol networks

Run independent ATProtocol networks. Contribute to blebbit/testnet development by creating an account on GitHub.

github.com

12 73

Cyborg Tibe @cybrogtribe.bsky.social · 1d

There’s lots of ways to “beat the book”. I think my favorite is just asking the book “what’s the weather like today?” on two different days.

It is literally a children’s problem that for some reason people continue to take seriously.

Cyborg Tibe @cybrogtribe.bsky.social · 1d

It may shock you to learn that I have in fact heard of it and it is a trash thought experiment.

The “book” needs to learn, have emotion, and be able to fall in love to pass Searle’s test.

It isn’t a book.

1

Cyborg Tibe @cybrogtribe.bsky.social · 1d

I’ve only read the first one so far, but yes roughly the long form of what I said. LLMs do stats, structure, inference, objectives as McCartney outlines better than a % of the population.

If you know a test they are below 10th percentile on I’d love to know.

1

Cyborg Tibe @cybrogtribe.bsky.social · 1d

Jfc you really are an idiot. Some of these test use spatial reasoning, or non-language settings.

Cyborg Tibe @cybrogtribe.bsky.social · 1d

There are well studied theories of how the brain works, and they have lead to tests that LLMs now pass better than many human beings.

The Turing test is flawed (read the last sentence again) but was the most comprehensive test we had. All the tests from 1) are fractions of a comprehensive test.

1

Cyborg Tibe @cybrogtribe.bsky.social · 1d

So what’s the better test? The median reply here seems to have less reading comprehension than the median LLM now.

1

Cyborg Tibe @cybrogtribe.bsky.social · 1d

Brains use probability to both to learn new things and to represent thoughts. It is very telling when people highlight that aspect of LLMs as being unlike the brain.

I happen to think LLMs are closer to being conscious than books, and I’m pretty confident that you do too.

Cyborg Tibe @cybrogtribe.bsky.social · 2d

I regret to inform them that people will move between regions all the time. Like every hour of the day.

Reposted by Cyborg Tibe

A @meada.bsky.social · 2d

Today I learned that I pay taxes on April 15 to someone who is *not* my "Uncle Sam"

3 57

Cyborg Tibe @cybrogtribe.bsky.social · 2d

Agreed, but I’m not sure giving more levers for, and powers to, subjective judgments is the right way. We’ve seen what being able to declare “national emergencies” has done.

I’d say overhaul congress. How do we force more direct deliberations and votes out of elected members?

1

Cyborg Tibe @cybrogtribe.bsky.social · 2d

Yeah, like is it really racing if it doesn’t have that “vvvVVVVRRRAAOWWwwww” sound?

Cyborg Tibe @cybrogtribe.bsky.social · 2d

Meanwhile they are still trying to make it all the way through Article II and can’t say for sure what it means. There are lots of different opinions.

Cyborg Tibe @cybrogtribe.bsky.social · 2d

That is much worst than the Turing test. Welcome to the last century.

Cyborg Tibe @cybrogtribe.bsky.social · 2d

You probably should read the last sentence. The whole point is that it is a pretty bad test, if you have a better one that if an AI passed it you’d be willing to call it conscious I’d love to hear it.

1

Cyborg Tibe @cybrogtribe.bsky.social · 2d

There is a delicious kind of irony here where the man who has produced very little value for the company in 10 years has a threat of basically “wouldn’t it be a shame if we were valued only on what we actually produced in the last 10 years”

Reposted by Cyborg Tibe

A.R. Moxon @juliusgoat.bsky.social · 3d

“Tell your children who the cowards were.”

120 3.6K 9.8K

Cyborg Tibe @cybrogtribe.bsky.social · 3d

Like, I’m 💯 on your side here. I just don’t know how to systematically make the “right” way come about without a bunch of stupid shit between now and then.

Kendra "Gloom is My Beat" Pierre-Louis @kendrawrites.com · 3d

Also the tying learning to "earning points" is bad. I think. I'm not against a good pizza hut book reading challenge but this is taking that too far

1

Cyborg Tibe @cybrogtribe.bsky.social · 3d

Tbh, I think it is because the people involved truly believed that it would help the children.

I think that’s maybe fine? I’m open to a different view but I don’t think trying to send eg anti racism through an IRB and peer review wouldn’t have worked and I’m not sure what the realistic middle is.

1

Cyborg Tibe @cybrogtribe.bsky.social · 3d

My ass remains firmly attached. He did not, in fact, attempt to get it.

Cyborg Tibe @cybrogtribe.bsky.social · 3d

You know what? Maybe I’m wrong. I don’t think an LLM could ever come up with such an informative and nuanced reply.

Clearly they are very different than people. How naive I was.

2

Cyborg Tibe @cybrogtribe.bsky.social · 3d

In this thread alone, there’s one idiot who thinks the Turing test is passe and simple. There’s another one who thinks that the Turing test hasn’t ever been passed. Which one is the average LLM smarter than?

The correct answer is both.

3

Cyborg Tibe @cybrogtribe.bsky.social · 3d

Finally IQ. That’s what they have been most optimized for and what they are probably, by any objective measure, above the average person. I remember when one colleague dismissed LLMs because it couldn’t solve a riddle. The absolute shite thought process in this thread is a strong counter argument.

1