With the OpenAI clownshow, there's been renewed media attention on the xrisk/"AI safety"/doomer nonsense. Personally, I've had a fresh wave of reporters asking me naive questions (as well as some contacts from old hands who are on top of how to handle ultra-rich man-children with god complexes). 🧵1/ - ThreadSky

emilymbender.bsky.social • 531 days ago

With the OpenAI clownshow, there's been renewed media attention on the xrisk/"AI safety"/doomer nonsense. Personally, I've had a fresh wave of reporters asking me naive questions (as well as some contacts from old hands who are on top of how to handle ultra-rich man-children with god complexes). 🧵1/

Comments

alanrew.bsky.social•530 days ago

This is a wonderful thread. I'm saving the first post to share with anyone who succumbs to this sort of nonsense.

alanau.bsky.social•531 days ago

Personally, I find it helpful to remember that language is a method for communicating ideas. The challenge, and currently non-machine-replicable part, is the idea generation part of the process.

I sometimes quip that a machine can win at chess but would never play chess unprompted.

oldmate.bsky.social•531 days ago

I used a similar argument the other day - the machines can beat us at chess, but they cannot make us play.

brianpenny.bsky.social•531 days ago

Amazing breakdown. Thank you!

I shared this entire thread on IG Threads as it’s the one microblogging social network you’re not on but is filled with AI proponents parroting OpenAI talking points.

doctecazoid.bsky.social•531 days ago

Say what you want about ChatGPT, OpenAI or its evangelists ... once Corporate America (#AmeriCo) gets its hands on it, they're gonna use it to maximize profits and squeeze the middle class out of existence. Barring unforseen circumstances (& hey, I admit I could be wrong!), it's a fait accompli.

coffeecannabisclub.bsky.social•531 days ago

You aren't wrong.

This is exactly what IS going to happen. The only question is the expediency in which it happens.

This coupled with the current rate of decent into a full blow fascist plutocracy...

Hold on to yer butts. Most people can't be inconvenienced into even trying to stop it.

pyrholidon.bsky.social•531 days ago

Oh we're well past 'once'.

https://www.usatoday.com/story/news/health/2023/11/19/unitedhealth-artificial-intelligence-denies-claims-in-error-lawsuit-says/71579822007/

doctecazoid.bsky.social•531 days ago

LOL I was just in the process of adding a link to this story as a reply to the tweet. Yeah, the tsunami is well underway.

emilymbender.bsky.social•531 days ago

As a quick reminder: AI doomerism is also #AIhype. The idea that synthetic text extruding machines are harbingers of AGI that is on the verge of combusting into consciousness and then turning on humanity is unscientific nonsense. 2/

missingthept.bsky.social•512 days ago

https://bsky.app/profile/missingthept.bsky.social/post/3kgbzkedqlh2t

rapieress.bsky.social•531 days ago

It reminds me of when ATMs came on the scene—they were going to eliminate the need for Tellers. Banking can be individual & complex, so we still have Tellers. Same with self-checkout. You still need some staff because there is always problems at the machine that need humans to fix.

sandysev.bsky.social•531 days ago

I actually think this generation of ai will enter into a death spiral at some point.

Based on Holyrood, they want to use it straight away. The irony is this might kill it.

benstokes.bsky.social•531 days ago

In other news... https://bsky.app/profile/unusualwhales.bsky.social/post/3kevearran62j

petronella.bsky.social•531 days ago

Hahahaha.

tillichfarris.bsky.social•531 days ago

100% correct. It's a case of mass hysteria.

rodgerkibble.bsky.social•531 days ago

I seriously think SF authors are partly to blame. Long before Terminator there were RUR, The Forbin Project etc. Where are the stories about people getting refused a job or receiving inappropriate medical treatment because of biased AI?

lfv-rhetrx.bsky.social•529 days ago

There have already been cases where MDs inside insurance companies have been rubber-stamping “AI-generated” denials of treatments; wasn’t there a ProPublica report of that last year?

sandysev.bsky.social•531 days ago

Oh. They are coming. I think they are just being suppressed.

natanael.bsky.social•531 days ago

https://bsky.app/profile/natanael.bsky.social/post/3keunllmh2y2a

Some aspects were predicted, like 1984's versificator. It's not an exact match but close enough, it was meant to take the act of creative work out of human hands and simultaneously keep the proles in check. Generating news, songs, movies, etc.

emilymbender.bsky.social•531 days ago

At the same time, it serves to suggest that the software is powerful, even magically so: if the "AI" could take over the world, it must be something amazing. 3/

emilymbender.bsky.social•531 days ago

Reporters working in this area need to be on their guard and not take the claims of the AI hype-mongers (doomer OR booster variety) at face value. It takes effort to reframe, but that effort is necessary and important. We all, but especially journalists, must resist the urge to be impressed: 4/

leojallenjr.bsky.social•531 days ago

In my opinion, the most impressive thing about any and all AI at this point is the coding architecture. All of the cute pictures are cool and interesting, but that's about it. The writings are also cool and all, but again - that's it. "Speed" is not "intelligence". Nor is it a sign of life.

leojallenjr.bsky.social•531 days ago

I keep trying to remind people that none of these things are anything without the people behind them. Don't fear AI. As with any and all tools, fear the people behind AI. A tool is only destructive when a human makes it so. What we are seeing with OpenAI is merely capital generating fluff.

marypcbuk.bsky.social•530 days ago

OpenAI’s needs were a great forcing function for getting kubernetes clusters in the cloud to scale with GPUs. I’d put that as one of their most useful outputs

sffcorgi.bsky.social•530 days ago

Irrelevantly, that bird really needs its beak trimmed.

emilymbender.bsky.social•531 days ago

As a case in point, here's a quick analysis of a recent Reuters piece. For those playing along at home, read it first and try to pick out the hype: 5/
https://www.reuters.com/technology/sam-altmans-ouster-openai-was-precipitated-by-letter-board-about-ai-breakthrough-2023-11-22/

emilymbender.bsky.social•531 days ago

The article starts with some breathless but vague reporting about an unpublished and completely unsubstantiated "discovery" and "[threat] to humanity". Will the body of the article provide actual evidence? (Of course not.) 6/

emilymbender.bsky.social•531 days ago

Remember, this is the same company whose Chief Scientist says that "ChatGPT just might be conscious (if you squint)" (and gets this remark platformed by MIT Tech Review, alas)

7/

marshray.bsky.social•531 days ago

I don’t think that’s the most likely scenario either, but do you have a basis for this claim?

olespaarmann.bsky.social•531 days ago

Models like GPT4 are great in meeting user’s expectations. But they don’t produce any new knowledge. With this amount of memorized “facts”, a human clearly would. LLMs are just extremely good statistical parrots and maybe compression is all the magic there is. https://arxiv.org/abs/2311.13110

olespaarmann.bsky.social•531 days ago

So no Skynet just yet 🦾

olespaarmann.bsky.social•531 days ago

In other words: Yes, GPT can write your sales email in 5 seconds and parrot knowledge from the web. But that maybe says more about these tasks than about GPT. It can not, however, create knowledge. It doesn’t UNDERSTAND the world, it is just good in describing what it already knows.

davgar51.bsky.social•530 days ago

How can we be certain that this isn’t exactly what our minds are doing to?
We do have higher functions as well but we learn very well from examples vs. abstract arguments (for eg).
I am neutral on this but when I see that statement it seems more like an assertion than an evidenced position.

olespaarmann.bsky.social•530 days ago

ML mimics what we know about the mind. But claiming that maybe our brains work like an LLM just because there are similarities (that researchers designed) puts the cart before the horse.

jerembluehouse.bsky.social•531 days ago

Chatgpt is little more than the equivalent of a nail gun. In skilled hands, it can accelerate good work.
In the wrong hands.... Someone is going to lose an eye or a hand.

drdarkstar.bsky.social•530 days ago

Define “understand” :)

genej.bsky.social•528 days ago

It is extremely hard to avoid anthropomorphizing ChatGPT when it ads eerily “empathetic” and “thoughtful” comments answering a prompt. Also humans are hard-wired to equate intelligence with quick responses (most IQ-related tests are timed) and ChatGPT responses are incredibly fast.

drdarkstar.bsky.social•528 days ago

Don’t disagree (although I don’t think I anthropomorphize it here, I don’t consider it sentient or intelligent or conscious) — but I don’t feel like that need be relevant to “understanding” text.

genej.bsky.social•528 days ago

We are definitely entering into the “uncanny valley” chatGPT equivalent. It should be relatively simple to train chatGPT to obfuscate any attempts to test it for sentience. We are entering “the brave new world”

olespaarmann.bsky.social•530 days ago

Here we quickly get to the “hard problem of consciousness” in the Philosophy of Mind. I would argue that we don’t know enough about our minds to fully define “understand”, but enough to know that LLMs don’t qualify. We can only look at capabilities, the qualia are opaque (both in humans and LLMS)

drdarkstar.bsky.social•529 days ago

I don’t think “understanding” requires consciousness at all, I feel like you are taking a very extreme and narrow view there.

It’s a bit circular (all the best things are) but to understand means to comprehend intended meaning…

drdarkstar.bsky.social•529 days ago

When I give gpt some shell commands and ask it to “turn this shell script into a Java class” and it can do that, it certainly feels like that qualifies for “understanding” in a real and meaningful (but not conscious) way. It has to unpack what the shell commands “mean” in order to translate them.

everythingism.bsky.social•531 days ago

It's quite noticeable lately how journalists -- even those writing for what you thought were reputable publications -- fall for idiotic narratives on issues that involve just a modest amount of complexity, but where a false/misleading storyline has been seeded.

jeffreyo.bsky.social•531 days ago

I don’t think the journalists are necessarily falling for anything. The media is a business, like anything else, and their business is selling eyeballs. The hype isn’t a bug of journalism; it’s a feature.

everythingism.bsky.social•531 days ago

There may be some element of that but in general I think it's more likely people are really being duped. The Associated Press for instance doesn't have much to gain from one article about OpenAI.

everythingism.bsky.social•531 days ago

Some other examples include the hysteria around "UAPs" and the idea that the pandemic started with a lab leak. In all of these cases there's some degree of uncertainty, but most experts would tell you the more sensational interpretation is far less likely.

everythingism.bsky.social•531 days ago

However those invested in these stories have created enough "fake experts" and false assumptions that a superficial plausibility is manufactured. It's quite an insidious trend and disturbing in what it says about journalism (these stories implicate the NYT, WA Post, Intercept, ProPublica + others).

joshrosenau.org•531 days ago

Definitely speaks to the collapse of a lot of media outlets, experienced hands laid off, remaining reporters not given time to develop expertise, and too few eyes in the newsroom to catch bad takes before the story is locked in. And outlets can’t compete for tough stories, just re-report hot takes.

everythingism.bsky.social•531 days ago

I'm also tempted to say that journalists have never been immune to being fooled. It's just easier now, partly because the media is tackling issues that involve more complexity and partly because bad actors now have the tools to more easily manufacture fake plausibility.

joshrosenau.org•531 days ago

Yeah, as with science, the self-correction comes from openness, from competition and camaraderie, and from a lot of eyes catching errors. There have always been bad reports, but enough good ones that the truth comes out. News is slimming down and losing resources as its enemies ramp up.

Comments

Posting Rules

Reply