Write Rebuttals to Articles Dismissing X-Risk

The media is ripe with articles making a mockery of existential risk.

Examples:

I think we should systematically write rebuttals to these. It should not be too hard, because most arguments are empty. It is a good exercise in rhetoric that can be systematized.
I am happy to be Project Lead, but will need to find volunteers as I won't have much time to work on it. I am happy to provide mentoring in rhetorical analysis, this is an opportunity for someone who wants to develop these skills.

Help Needed
(Note that these roles do not have to be taken by different people).

A researcher: Tasked with finding and linking articles mocking x-risk.
A rhetorical analyst: Tasked with analysing the rhetoric of the articles, identifying individual arguments, rhetorical fallacies, and verifying facts (although these articles usually do not rely on facts).
A writer: Putting together a well-written and engaging rebuttal
A reviewer: Tasked with reviewing the rebuttal and providing constructive feedback

NOEMA

The Illusion Of AI’s Existential Risk | NOEMA

Focusing on the prospect of human extinction by AI in the distant future may prevent us from addressing AI’s disruptive dangers to society today.

AI will never threaten humans, says top Meta scientist

Artificial intelligence is still dumber than cats, says pioneer Yann LeCun, so worries over existential risks are ‘premature’

Stop talking about tomorrow’s AI doomsday when AI poses risks today

Talk of artificial intelligence destroying humanity plays into the tech companies’ agenda, and hinders effective regulation of the societal harms AI is causing right now.

dead-brownOP•2/19/24, 11:15 AM

Project Tracker: https://maximolog.notion.site/Write-Rebuttals-to-Articles-Dismissing-X-Risk-238504f4171c4ada8b5df64120a57b7a?pvs=4

Maxime Fournes's Notion on Notion

Write Rebuttals to Articles Dismissing X-Risk | Notion

Description

hurt-tomato•2/19/24, 12:04 PM

This is an amazing idea!

hurt-tomato•2/19/24, 3:30 PM

hit da brakes

colossal-harlequin•2/19/24, 6:57 PM

The way I've seen the vehicle metaphor used before is that commuters are on a bus in the fog, it's known the road is near cliffs, and people are earnestly discussing how late for work the fog is likely to make them.

But yes, emphasizing that so-called "mundane" risks are also of concern and that real solutions overlap considerably is less antagonistic. Ethics vs safety is largely a false dichotomy.

moderate-tomato•2/19/24, 9:49 PM

I could do some work on this. Following up with a call with Maxime, where could they be published? I suppose it doesn't matter as long as people are seeing the "article" (whether it's a blogpost or twitter thread).

Either way, I suspect it would be good to have it be more advocacy style / call to actiony (ex: https://act.represent.us/sign/are-presidential-primaries-rigged/, although that example is a bit too click-baity for my taste).

RepresentUs

Are the presidential primaries rigged?

moderate-tomato•2/19/24, 10:10 PM

Although, alternatively, are there more recent articles criticizing x-risk concerns? It might be better to rebut an article fresh in people's minds than bring up an old one.

dead-brownOP•2/20/24, 12:49 PM

You correctly noted that the publication platforms are key for the success of this project. I don't think PauseAI has a blog actually ( @Joep Meindertsma ? Should we have a blog?) so we need to think hard on that point. Twitter thread would be a good starting point, and I also like that it forces us to be concise. Sounds like a low hanging fruit

dead-brownOP•2/20/24, 12:51 PM

As for the recency, yes I would argue that it matters a lot, and an important part of this project is to identify such articles as soon as they come out. Especially when they are high visibility

dead-brownOP•2/20/24, 12:58 PM

found this potential candidate after a very quick research, plublished last month: https://www.scientificamerican.com/article/ai-survey-exaggerates-apocalyptic-risks/

Scientific American

AI Survey Exaggerates Apocalyptic Risks

A speculative survey about AI’s future may have been biased toward an alarmist perspective

Ddead-brown found this potential candidate after a very quick research, plublished last mont...

hurt-tomato•2/20/24, 3:30 PM

That might warrant a response, it is biased. Nirit-weiss Blad is one of the most toxic people on twitter - they really have an axe to grid with EA.

There are few points to really counter, though. The article itself mostly consists of hand-wavy criticism, trying to paint EA as some sort of evil big money cult.

Some points that can be countered:

"I see they’re still not including conferences that think about ethics and AI explicitly". The researchers only tried to include researchers who are deeply familiar with the technology itself - people who have proven themselves to show they actually understand the technology.

But I think the article is mostly fair?

dead-brownOP•2/21/24, 6:02 PM

https://observer.com/2024/02/metas-a-i-chief-yann-lecun-explains-why-a-house-cat-is-smarter-than-the-best-a-i/

Observersissiobserver

Meta’s A.I. Chief Yann LeCun Explains Why a House Cat Is Smarter Th...

“We are missing something conceptually big to get machines to be as intelligent as animals and humans.”

dead-brownOP•2/21/24, 6:28 PM

This is prime material for debunking

dead-brownOP•2/26/24, 9:58 AM

OK let's get this project started. As a start, I think we should use twitter as a platform for publishing these rebuttals. The format can be short and snappy, like a list of 10 points maximum that are wrong about the article. @Joep Meindertsma would it be OK to use the PauseAI Twitter account for this?
@bolun would you have time this week to start a rhetorical analysis of one of the two articles above (either the observer one, or the politico one submitted by Squink)?

dead-brownOP•2/26/24, 10:00 AM

In order for multiple people to be able to collaborate, I would suggest to create a google docs, paste the text from the article, and start commenting on it to isolate the fallacies, dubious facts, etc

hurt-tomato•2/26/24, 12:16 PM

I'll gladly post these to twitter!

dead-brownOP•2/26/24, 12:30 PM

https://docs.google.com/document/d/1NoR5n4KQnWWQKwhkpJ0gCjUg4uivy-YaGRSmLnrOsUY/edit?usp=sharing

Google Docs

Politico - AI doomsayers funded by billionaires ramp up lobbying

dead-brownOP•2/26/24, 12:32 PM

https://docs.google.com/document/d/1yagJRs75DyMOPX8NXZvfUFSpfLDZemcutMj1tr8hsQs/edit?usp=sharing

Google Docs

Observer - Meta’s A.I. Chief Yann LeCun Explains Why a House Cat Is...

sacred-rose•2/27/24, 12:54 AM

This is not an article but possibly a lot of people will read the comments https://www.reddit.com/r/dankmemes/comments/1aymi0x/not_what_they_expected/

From the dankmemes community on Reddit: Not what they expected

Explore this post and more from the dankmemes community

Ddead-brown OK let's get this project started. As a start, I think we should use twitter as ...

moderate-tomato•2/27/24, 9:28 PM

Sure, I'll try to finish a draft by ~~Saturday~~ Sunday.

Mmoderate-tomato Sure, I'll try to finish a draft by ~~Saturday~~ Sunday.

dead-brownOP•2/28/24, 10:36 AM

Awesome! I'm on holiday right now but would be available on Friday if needed

moderate-tomato•3/5/24, 3:59 AM

I made a few comments on the Observer & Politico articles, focusing on factual errors. The tone of the articles isn't what I'd prefer, but there isn't a clear cut way to rebut that.

moderate-tomato•3/5/24, 4:03 AM

They're brief -- I'd like suggestions as to what to look for.

Mmoderate-tomato I made a few comments on the Observer & Politico articles, focusing on factual e...

dead-brownOP•3/5/24, 9:33 AM

Just reviewed your comments on the observer article. This is great, there is definitely enough material to write a 5 to 10 points twitter post about it

dead-brownOP•3/5/24, 9:34 AM

Let's think about a format. Here is what I suggest:

first tweet is introductory, with some kind of title. "Everything wrong with [...]. This is an example of disinformation propagated by disingeneous actors with vested interests, putting the world in danger for their short term profit. [(feel free to change that but let's not mince our words]
each subsequent tweet is a quote from the article (as short as possible, a rephrasing if the quote is too long) followed by our take on it
final tweet concludes that the article has zero content and is pure noise, at best ridiculously naive, at worst intentional propaganda

dead-brownOP•3/5/24, 9:45 AM

In any case, amazing work @bolun

colossal-harlequin•3/5/24, 7:03 PM

With the Politico article, the headline is clickbaity though true. Other than that I don't think the reporting appears biased either way, that our arguments are with Suresh Venkatasubramanian not the article as a whole.

Additional arguments against Suresh that supplement those already provided by CAIS et al are of course worth making.

(Apologies for comments here. Some issue in adding them to the doc itself, probably me being dumb.)

Write Rebuttals to Articles Dismissing X-Risk

Similar Threads

Similar Threads

Similar Threads