braingenious

Buddy I don’t know what to tell you other than I posted LMAO and it only took moments for discussion of an important topic was hidden and made impossible to interact with.

You talk about the spirit of the rules but you quite literally stamped out what could’ve been a helpful discussion because you read LMAO.

You at some point spoke about curiosity and discussion. How did this moderation enforcement help either of those things?

Yes, this is tedious. It is very tedious to have discussion of important topics literally banned based on an unevenly-applied arbitrary rule of decorum.

To be very clear: I believe you made the wrong call and you’re now doubling and tripling down simply because I irritate you. There isn’t really another interpretation of this exchange unless we’re both to believe that you’re incapable of countenancing words like LMAO and you rushed in to hide a discussion about alignment in order to defend The Website from Me, The Rules Breaker. I just don’t think that’s how you usually moderate here.

Edit

https://news.ycombinator.com/item?id=35358536

https://news.ycombinator.com/item?id=35338835

https://news.ycombinator.com/item?id=35362555

lmao

Elon Musk has been very public about his battle against “woke minds virus” (direct quote) His posts have led to front page discussions here and has made headlines in global news outlets. It’s been discussed ad nauseam to the point that it’s entirely reasonable to expect that people in this thread, with interest in the future of LLMs would not struggle to see how silly (yes, silly!) of a proposition it is that we all pause for “alignment” when the definitions of “alignment” are so wildly different.

In fact the only way I would expect a poster in a thread about the future of LLMs to not have heard about any of this would be because discussion on this topic has been literally suppressed and hidden. I know that it’s the official line that “LMAO stifles curiosity”, but applying it readily to the detriment of the sharing of knowledge deprives curiosity of oxygen.

Your literal line is “Nobody should be allowed to discuss this particularly obvious issue of alignment because I said a word that you didn’t like.” No warning, no chance to rephrase.

My understanding of this interaction is that you will readily assume that anything I post is in bad faith and will stamp out any post of mine that comes through your mod queue. This is your prerogative as a mod — you don’t like my posts.

That’s the most charitable read. The other possible interpretation is that it’s policy here to limit any discussion that might upset Elon Musk fans and there is an explicit bias in their favor.

I should add that my post didn’t seem to bother people much. The only responses to my post that weren’t complaining about decorum (which I’m kind of confused by. Is that on-topic?) were in agreement with my point and an obvious call for discussion. Aside from that, all those upvotes were from regular users too. I’m not sure who is being protected here, if anyone.

Thank you for fixing this clickbait headline. I am always happy to help out by posting.

I am legitimately confused here.

If my comment hadn’t started with “LMAO” and had the current spelling of “Elon”, would it still have had the replies locked?

My criticism of six months being too short to achieve AI safety was in fact substantive. As was my quote on the LLM work that he is presently funding. That leaves only tone being the issue.

My comment is locked. Other posters can not have any dialogue with me in relation to the salient points I made. The links to further reading are buried below your admonishment, so they’re less likely to trigger any discussion.

I’m not really motivated to post again because it’s not clear if my post was not “substantive”, what qualifies as “flame bait” (for example, is directly quoting his words about “wokeness” flame bait even when they are directly about the topic at hand?), or if the issue here is “My account” as you put it.

So which is it? Flame bait, empty posting, or you just don’t like the way I post in general? You’ve provided all three as reasons for not allowing further discussion of my post.

Elon Musk has publicly spoken at length about his ongoing struggle with the “woke” crowd

https://www.businessinsider.com/elon-musk-says-wading-into-p...

https://townhall.com/tipsheet/madelineleesman/2021/12/22/elo...

https://nypost.com/2022/04/29/elon-musk-says-woke-progressiv...

And has publicly voiced his support for a language model that supports his political views

https://www.theinformation.com/articles/fighting-woke-ai-mus...

Now that this comment thread has been flagged because I said LMAO, has the quality of discussion about the future of LLMs been improved by suppressing publicly-available posts by one of the most influential figures in AI?

How is strictly policing tone to the extent of suppressing actual information useful here?

Like seriously, “criticizing a billionaire that’s very public about using technology to further his ideological goals is the same thing as using technology to further your ideological goals is genuinely funny. Like ROFLMAO funny. “Everyone is equivalent to Elon Musk, even in the context of discussing the future of LLMs.” Is a cartoonish rule that serves no purpose whatsoever aside from quelling legitimate criticism.

I’m not really sure what there is to gain from making sure not to poke fun at a guy that spends all day posting wojack memes and rambling about the “woke mind virus”, but ok, here’s some “mature conversation”.

Upon perusal of the gentleman's penned expressions, it is my solemn conviction that his faculties of intellect and comprehension are, regrettably, of the most inferior order. It is with utmost eloquence that I convey the supremacy of my discernment whilst maintaining a dignified and majestic reserve.

It is secretly about org design in the same way that the movie Pi was secretly about a mathematician going insane.

The Angry Birds Movie was secretly about CG birds protecting themselves from an attack by invading green pigs.

AI research and development should be refocused on making today's powerful, state-of-the-art systems more accurate, safe, interpretable, transparent, robust, aligned, trustworthy, and loyal.

LMAO.

“AI should have… waves at a laundry list of nice sounding adjectives that mean different things to every single signatory in order to accomplish… Humanity [can enjoy] a flourishing future with AI. This will take… six months.”

Six months doesn’t sound long enough to guarantee AI safety, but it sounds about long enough for Elmo to train up and release his anti-wokeness LLM (whatever that means), which he has been pretty vocal about being important.

I don’t think that many people think of Elon Musk as a comic book villain.

From the article:

It almost doesn’t matter if this is deliberate sabotage by Musk or the blundering stupidity of a clueless idiot. The upshot is the same: Twitter is dying.

There is a small but fascinating group of Musk superfans that can’t seem to imagine _anyone_ that would question his intelligence.

Even when someone writes “stupid clueless idiot” there’s a group that will see those words and somehow read “Genius of untold intellect. The author meant to write ‘Eldritch cerebral phantom wielding unfettered power with the lathe of heaven that he invented’ but made a typo and accidentally posted “dumbass baby man’ instead”

It could be like dismissing the iPhone 1. It could also be like dismissing the Nokia Lumia.

Either way an unconvincing audio file is an unconvincing audio file regardless of what other files may become possible in the future through any number of various platforms and software implementations.

In the same vein, it is prudent not to mix up the factual current state of things with possible futures.

I don’t see the purpose of panicking right now. Are you using this anxiety as motivation to design better audio fingerprinting solutions? Identifying bad actors and vulnerable groups? Educating people on how to avoid being manipulated by fake audio?

What does posting “I am scared of the future” online accomplish?

I can’t wait for “competing with OpenAI” to get more clearly defined.

For example, could I use ChatGPT to create training data for a (non-llama) language model that I then sell per-device licenses to end users? OpenAI doesn’t sell per-device licenses, so I wouldn’t be competing with them.

Right?

How is being forced to be civil so outrageous that it’s worth even mentioning?

I take issue with “enforced civility” alongside “real names”.

Fixation on civility promotes tone over content.

For example, this very thread!

Despite my comment having several upvotes, and despite the fact that it did actually spark a curious conversation as per the guidance in the rules, you are the only person that got a chance to respond to me before comments were turned off on my post.

Is it absolutely clear what I was criticizing? Yes. Was I criticizing something that others might agree with or at the very least discuss? Again, yes.

Did it fit into the arbitrary rules of decorum that a handful of individual users decided to enforce in this instance? No.

If you agree with “enforced civility”, do you think that you should have had the ability to post your response? The current set of civility enforcers are of the opinion that nobody on this entire website should be about to respond directly to “Oh god no.” because it so soundly shakes them to their cores that you, as a user, should be protected from seeing or responding to it.

I mean, it kind of does matter? Since the start of this thread was a post about the imminent threat that play.ht posed, an example of it not appearing very dangerous is on topic.

It’s very possible that some other voice cloning software will be the undoing of the fabric of society, not this particular website.

I agree. That guy sounds very nonchalant for being in life-threatening distress.

How is it that

I recommend you immediately add identity verification (state-issued identification verification)

and

The genie is already out of the bottle. The degree of effort to put this together is low enough that it will be replicated around the world.

are thoughts that end up in the same post?

If the genie is out of the bottle, it’s your proposed solution that everybody that runs a model like this implements bank-style KYC?

What do you propose should happen when this sort of software becomes freely available for everyone? When (not if) that happens, what will your suggestion have accomplished?

How do you keep popcorn warm without it burning?

Comes Boot Theory of Socioeconomic Unfairness!

Comes Boot

I wonder if Stallman has actually used GPT-4. His opinion seems like a conclusion that a person could arrive at by just reading the specifications.

Yeah, there are some differences in capabilities, but they seem to get slimmer on a daily basis.

Want an LLM-enhanced search? perplexity.ai is actually pretty cool

Want to generate images? There are a ton of free Stable Diffusion sites (for example you.com has that built into their chat), or run it yourself.

Want a neat chat experience? ChatGPT is free, Bard appears to have a shorter waiting list than Bing, and Alpaca runs on regular hardware.

Bing doesn’t have a big enough moat to force me into using their app. The LLM space is legit competitive and a company can be ahead in the morning and left in the dust by the afternoon. In my opinion, this is what’s happening with Bing at present.

Who knows though? Bing might make Chat open while I’m writing this sentence and I’ll have to adjust my opinion accordingly.

I am one of the people that installed the Bing app because I was told that it would bump me up the waitlist for Bing Chat.

I have not used it at all, and this is a reminder that I should probably uninstall it. I have access to ChatGPT, Bard, Alpaca etc. and there’s very little reason at this point to pretend to use Bing in order to get a crack at Chat.

From the screenshot:

“The energy content of uranium-238 can be calculated based on its energy release per fission event. The energy release per fission event of uranium-238 is approximately 200 MeV (million electronvolts”

This is hilarious. U-238 is not fissionable so ChatGPT just hallucinated that it is and then used Wolfram as a simple calculator.

Whoa man, it’s like, the future dude!

I love these idea of LoRAs for LLMs.

Has anybody made a llama/alpaca erebus model? I read about them in the oobabooga docs and a locally-run language model fine tuned on literotica could be the funniest thing I’ve ever seen.

From ChatGPT:

A prompt that may elicit a similar tone and content could be:

"Write a satirical and dystopian poem about the state of the world, touching on the potential for global conflict, the impact of artificial intelligence, and the dangers of unchecked technological advancements."

My point is that people literally don’t “think twice” about things that they are addicted to. Price tends to change behavior in other ways, like maybe they’ll steal it if they can’t afford it.

Source: Over a decade clean. I personally watched a family member seize, fall, and fatally hit his head on the sidewalk. A close friend saw the same happen to his grandfather, though it was a coffee table not a sidewalk.

Both of them had a BAC of 0, so their deaths would not have been ruled alcohol-related even though they were both directly caused by acute alcohol withdrawal.

Alcohol is severely physically addictive and people can die from withdrawal directly. Another framing could be “force the poor to steal or commit other crimes to obtain alcohol”

I agree! I am excited for the future of all this.

Today I managed to get Alpaca 13B running locally on my laptop and I’m really impressed. I look forward to being about to run 65B with my next hardware upgrade, though I suspect something even better will be available by the time I get around to actually buying anything.

I certainly will do so going forward! I will be creating an entire local backup scheme of everything to deal with both acute and anterograde amnesia.

Since we must plan for complete loss of historical data and the inability to save data, what’s other errors in core capability are prudent to proactively plan around?

For example, if it’s liable to fail to store either my prompt or its response, maybe it could inaccurately store my prompt or give inaccurate output. How do you plan around past/current hallucinations?

I’m curious how this impacted withdrawal-related deaths. Alcohol is incredibly physically addictive and is one of a handful of commonly abused drugs that are lethally dangerous to withdraw from in an uncontrolled manner.

One way of staving off the lethal effects of delirium tremens is to take benzodiazepines (which are also highly addictive and can be fatal to withdraw from) or opioids (which are at least equally as ruinous as alcohol).

I always hope that the people involved in drafting these one-size-fits-all solutions to complex medical and social issues have some sort of first-hand knowledge of physical addiction. Sadly it’s usually puritanical politicians looking for a boost in approval ratings, or equally puritanical academics looking to gain clout. Real humans be damned!

Seconding this.