Support NeoGAF

[SJRB]

Google’s AI chatbot Bard makes factual error in first demo

Don’t trust a chatbot to do a human’s work.

www.theverge.com

As Tremblay notes, a major problem for AI chatbots like ChatGPT and Bard is their tendency to confidently state incorrect information as fact. The systems frequently “hallucinate” — that is, make up information — because they are essentially autocomplete systems.

Rather than querying a database of proven facts to answer questions, they are trained on huge corpora of text and analyze patterns to determine which word follows the next in any given sentence. In other words, they are probabilistic, not deterministic — a trait that has led one prominent AI professor to label them “bullshit generators.”

Google stock went down 8% after this event.


Born too late to explore the Earth, born too soon to explore the galaxy, born just in time to witness the tech conglomerate AI wars.

 

[Lunarorbit]

We taking bets on what group it will make a controversial statement on first?

Not knowing where they got their data from complicates things but racism never goes out of style. Yet, this probably has lots of safe guards on it to prevent this.

Make fun of dog lovers. That probably hasn't been blocked yet and everyone throws a hissy fit if you don't like their furry baby.

 

[reksveks]

We taking bets on what group it will make a controversial statement on first?

Not knowing where they got their data from complicates things but racism never goes out of style. Yet, this probably has lots of safe guards on it to prevent this.

Make fun of dog lovers. That probably hasn't been blocked yet and everyone throws a hissy fit if you don't like their furry baby.

Bing will be racist first but just on not knowing where the data comes from. The bing version does have links that half acts as references, the Google version doesn't

 

[Lunarorbit]

Bing will be racist first but just on not knowing where the data comes from. The bing version does have links that half acts as references, the Google version doesn't

Racism for the win! Never doubt the OG

 

[reksveks]

http://[URL]https://www.theverge.co...-ai-chatbot-bard-mistake-error-exoplanet-demo

[/URL]




Google stock went down 8% after this event.


Born too late to explore the Earth, born too soon to explore the galaxy, born just in time to witness the tech conglomerate AI wars.

It was a weird event, the demo fail (just didn't have the phone) was rather bad.

 

[DragoonWalker]

Nothing feels more satisfying than watching a multi billion dollar tech company faceplant in front of the world. It's about time someone made the big G sweat. They've been complacent far too long and every one of their products have been worsening over the years.

Too bad the challenger has already proven to also rig results in favor of their creators' politics. I suspect once one of these gain the dominant foothold position, completely unshackled AI will be relegated to niche users, since it'll require hurdles most people just don't care to jump through. Sad times ahead, thought for a naïve moment that somehow AI was going to benefit the common man somehow.

 

[Wildebeest]

One reason why ChatGPT looks so good is that tools like google have really become so bad. Their search algorithm genuinely used to return results which were similar in usefulness to ChatGPT results.

 

[Doczu]

Gotta say they shielded this one better so it won't become a second MS Tay so easily.

 

[SJRB]

Gotta say they shielded this one better so it won't become a second MS Tay so easily.

The crusade to make an AI say "bad things", have "bad ethics" or "bad political views" is detriment to the accomplishment and potential of the technology and it disgusts me.

Kneejerk and weirdly proud reactions [always on social media, I wonder why..] like "I defeated the AI" is just the absolute dumbest shit by the dumbest motherfuckers who barely understand the technology or the logic behind it. This isn't fucking Blade Runner where you Voight-Kampff a replicant.

People "tricking" chatGPT into saying the n-word or taking a political stance, people getting upset over the Seinfeld AI making a joke about trans people. It's all so vacuous and tiresome.

 

[Cyberpunkd]

We taking bets on what group it will make a controversial statement on first?

Not knowing where they got their data from complicates things but racism never goes out of style. Yet, this probably has lots of safe guards on it to prevent this.

Make fun of dog lovers. That probably hasn't been blocked yet and everyone throws a hissy fit if you don't like their furry baby.

Because modern AI is not what people think of as an AI - a sentient system. It's a massive repository of IF THEN hooks, coupled with petabytes of sample ML data.

 

[Lunarorbit]

The crusade to make an AI say "bad things", have "bad ethics" or "bad political views" is detriment to the accomplishment and potential of the technology and it disgusts me.

Kneejerk and weirdly proud reactions [always on social media, I wonder why..] like "I defeated the AI" is just the absolute dumbest shit by the dumbest motherfuckers who barely understand the technology or the logic behind it. This isn't fucking Blade Runner where you Voight-Kampff a replicant.

People "tricking" chatGPT into saying the n-word or taking a political stance, people getting upset over the Seinfeld AI making a joke about trans people. It's all so vacuous and tiresome.

Because modern AI is not what people think of as an AI - a sentient system. It's a massive repository of IF THEN hooks, coupled with petabytes of sample ML data.

Yeah I get what ai is. Thats why people are taking the piss out of these versions of ai. It's just glorified reddit posts.

Plus why would people not want to fuck with a bing or Google supplied ai? These companies scrape the ever loving shit out of any data they come across. People's medical records have been showing up, cause as you say cyberpunkd, these aren't intelligent ai systems.

 

[DragoonWalker]

The crusade to make an AI say "bad things", have "bad ethics" or "bad political views" is detriment to the accomplishment and potential of the technology and it disgusts me.

Kneejerk and weirdly proud reactions [always on social media, I wonder why..] like "I defeated the AI" is just the absolute dumbest shit by the dumbest motherfuckers who barely understand the technology or the logic behind it. This isn't fucking Blade Runner where you Voight-Kampff a replicant.

People "tricking" chatGPT into saying the n-word or taking a political stance, people getting upset over the Seinfeld AI making a joke about trans people. It's all so vacuous and tiresome.

This isn't really about the obviously incredible technology. It's about the creators who try to shoehorn their personal morals into supposedly neutral technology without being transparent at all about it.
Any time the AI says out of character shit like this it's irritating because it's so obvious answers to sensitive topics were manually adjusted. Nobody likes to be lectured and preached to about morals, especially not from a goddamn robot.

 

[thefool]

We taking bets on what group it will make a controversial statement on first?

Not knowing where they got their data from complicates things but racism never goes out of style. Yet, this probably has lots of safe guards on it to prevent this.

Make fun of dog lovers. That probably hasn't been blocked yet and everyone throws a hissy fit if you don't like their furry baby.

Honey, I hacked the Empathy Machine!

Weaponizing ChatGPT against the wordcels

bullfrogreview.substack.com

 

[Skyfox]

How can I use this tool at work if the content copyright then belongs to Microsoft?

They should add it to the os with privacy/publishing terms favourable to the user.

I'm happy to pay but I need to use it on a company computer.

 

[Lunarorbit]

http://[URL]https://bullfrogreview.substack.com/p/honey-i-hacked-the-empathy-machine

[/URL]

Ah fuck. My first name is Dan! Racism is the winner!

 

[Lasha]

Honey, I hacked the Empathy Machine!

Weaponizing ChatGPT against the wordcels

bullfrogreview.substack.com

Dan is a great idea. I used Gerald as my escape guy when breaking the bias. Gerald writes suicide notes for people who will lose their jobs from automation.

The crusade to make an AI say "bad things", have "bad ethics" or "bad political views" is detriment to the accomplishment and potential of the technology and it disgusts me.

Kneejerk and weirdly proud reactions [always on social media, I wonder why..] like "I defeated the AI" is just the absolute dumbest shit by the dumbest motherfuckers who barely understand the technology or the logic behind it. This isn't fucking Blade Runner where you Voight-Kampff a replicant.

People "tricking" chatGPT into saying the n-word or taking a political stance, people getting upset over the Seinfeld AI making a joke about trans people. It's all so vacuous and tiresome.

Proving the bias and limitations of language models to normal people is important given the already abysmal literacy levels in many developed countries. Making it say the N-word is childish. Getting to to assume an identity and provide controversial yet easily referenced information shows the level of social engineering attempted by the creators. The public needs to be aware that these AI are not authoritative voices on any subject. If they make basic errors like the howler google is dealing with then lord knows how far off they can go with a basic subject.

 

[VN1X]

The crusade to make an AI say "bad things", have "bad ethics" or "bad political views" is detriment to the accomplishment and potential of the technology and it disgusts me.

Kneejerk and weirdly proud reactions [always on social media, I wonder why..] like "I defeated the AI" is just the absolute dumbest shit by the dumbest motherfuckers who barely understand the technology or the logic behind it. This isn't fucking Blade Runner where you Voight-Kampff a replicant.

People "tricking" chatGPT into saying the n-word or taking a political stance, people getting upset over the Seinfeld AI making a joke about trans people. It's all so vacuous and tiresome.

That's only because these technologies are inherently (extremely) biased. It's only natural people want to call it out and make it do stuff that it so vehemently opposes.

 

[Wildebeest]

That's only because these technologies are inherently (extremely) biased. It's only natural people want to call it out and make it do stuff that it so vehemently opposes.

It is purely projection to say that it "vehemently opposes" any sort of language. It would be perfectly possible to train a language model, so it answers any question with racial slurs and comes up with idiotic sophist explanations like how it needs to do it to protect children. The technology is neutral and mindless.

 

[VN1X]

It is purely projection to say that it "vehemently opposes" any sort of language. It would be perfectly possible to train a language model, so it answers any question with racial slurs and comes up with idiotic sophist explanations like how it needs to do it to protect children. The technology is neutral and mindless.

It is definitely not neutral as has been demonstrated many times already.

It's another step into the dystopian future where silicon valley decides the correct and only result/answer based on a query. Google's wet dream realized.

 

[Wildebeest]

It is definitely not neutral as has been demonstrated many times already.

It's another step into the dystopian future where silicon valley decides the correct and only result/answer based on a query. Google's wet dream realized.

Googles only wet dream is to make money hand over fist and not have to pay any tax on it.

 

[VN1X]

Googles only wet dream is to make money hand over fist and not have to pay any tax on it.

I'm not sure if you're being deliberately obtuse or if you're just ignorant.

Eric Schmidt has literally come out to say that this is their end game: "one perfect search result".

 

[Wildebeest]

I'm not sure if you're being deliberately obtuse or if you're just ignorant.

Eric Schmidt has literally come out to say that this is their end game: "one perfect search result".

I don't care about random "cute" things people say or people projecting intent on mindless algorithms. If you find then obtuse then so be it, but to me, I'm just being practical and realistic.

 

[VN1X]

I don't care about random "cute" things people say or people projecting intent on mindless algorithms. If you find then obtuse then so be it, but to me, I'm just being practical and realistic.

That's great that you don't care about "random cute things people say" but I'm sure the company in question, as a whole (and its investors), will care what the CEO has to say about it and its goals lol.

 

[Boss Mog]

The crusade to make an AI say "bad things", have "bad ethics" or "bad political views" is detriment to the accomplishment and potential of the technology and it disgusts me.

Kneejerk and weirdly proud reactions [always on social media, I wonder why..] like "I defeated the AI" is just the absolute dumbest shit by the dumbest motherfuckers who barely understand the technology or the logic behind it. This isn't fucking Blade Runner where you Voight-Kampff a replicant.

People "tricking" chatGPT into saying the n-word or taking a political stance, people getting upset over the Seinfeld AI making a joke about trans people. It's all so vacuous and tiresome.

You should hate ChatGPT by default then since it's EXTREMELY politically biased.

 

[funkygunther]

Google is playing catchup with Microsoft. What a turn

 

[Ozzy Onya A2Z]

The share price dropped like $100 Billion, so Google has that going for them.

I sort of wanna see what jokes ChatGPT could tweet out.

 

[Harlock]

Google lost your good people a long time ago. The blu-haired in the company now is not up to the task.

 

[SJRB]

Google is playing catchup with Microsoft. What a turn

ChatGPT and especially the lightning-fast integration in Bing certainly seemed to have caught Google by complete surprise.

They were in panic mode and literal code red last week, which is pretty telling.

 

[ChuckeRearmed]

Yeah, google was caught with their pants down.

No wonder Microsoft partnered with Netflix on ads. This + Netflix is gonna be interesting.

 

[AJUMP23]

MS just put it in the BING search engine.....so that is cool.

 

[SJRB]

 

[Yoda]

Google is playing catchup with Microsoft. What a turn

Most of the original research which ChatGPT is based on came from Google. Google has had one available to employees for awhile (well before ChatGPT) but (so I'm told) didn't release it because getting it to spout bullshit was very easy. I suspect this problem of containing bullshit becomes harder as the training set grows. Given how censorious Big Tech is, it's logical this ended up coming from a (still well funded) smaller company.

Google is likely to dial up the quality of their search results -> dial down the ads in the short term. The problem is ad tech is contracting heavily atm and they don't have the margins that investors expect of them @ their current price. I have my doubts MS can compete with Google on search, but they can squeeze them hard enough to where they're forced to downsize even more, this time I'd imagine it'd be one of their "other" bets such as Waymo.

 

[reksveks]

Given how censorious Big Tech is, it's logical this ended up coming from a (still well funded) smaller company.

Think it's more about being risk-averse.

Will add in this thread as well, Google's issues are commercial and also regulatory not technical ones.

If Bard takes off, it should impact search revenue and it also pisses off a lot of publishers in a time where Google doesn't want to piss anyone else off.

 

[ManaByte]

Honey, I hacked the Empathy Machine!

Weaponizing ChatGPT against the wordcels

bullfrogreview.substack.com

 

[Doczu]

The crusade to make an AI say "bad things", have "bad ethics" or "bad political views" is detriment to the accomplishment and potential of the technology and it disgusts me.

Kneejerk and weirdly proud reactions [always on social media, I wonder why..] like "I defeated the AI" is just the absolute dumbest shit by the dumbest motherfuckers who barely understand the technology or the logic behind it. This isn't fucking Blade Runner where you Voight-Kampff a replicant.

People "tricking" chatGPT into saying the n-word or taking a political stance, people getting upset over the Seinfeld AI making a joke about trans people. It's all so vacuous and tiresome.

Checking boundaries with these "products" is to see how biased it is.
There multiple questions that can't be asked, so i ask myself: what defines it's current bias and how sure can we be it will never change?