5 ms·

Google's Ban on Election Queries in Gemini AI Sparks Debate

30 points by pg_1234 3y ago

oldpersonintx 3y ago
Google is mortified by the prospect of Gemini veering off away from blanket endorsement for Establishment Democrats
sandspar 3y ago
This seems more like an "I'm not a threat" signal for governments than a "you shouldn't learn the bad-bad words" nanny scolding to consumers. "Preventing misinformation" and "upholding democratic values" are just the ostensible reasons for this, the deeper reason is to avoid being regulated by angry congressmen. Perhaps they have deeper reasons as well, layer upon layer of bullshit.
- hnfong 3y ago
  > This seems more like an "I'm not a threat" signal for governments In other places, this would be called "self-censorship".
aaa_aaa 3y ago
Perhaps they want to hide their extreme bias which would be revealed in a comical way.
politelemon 3y ago
> Sparks Debate I did not see any reference to these debates mentioned in the title. Is this just a euphemism for something the author wants to question or bring attention to without blatantly giving an opinion.
- CapsAdmin 3y ago
  I'm sure if pressed, the author tweeted about it from one side and got a response arguing the other side.
nolok 3y ago
It's a very useless debate. Fact is, those are LLM, not "AI" the way the general populace understand that term. Fact is, that means they just give whatever answer they can derive from their training corpus, while people think they give facts. Fact is, you do not want to be liable for your half research / half pr product being manipulated into giving wrong or distorted facts regarding one of the most controversial and pivotal election, or worse not needing manipulation just straight up hallucinating. "Prompt engineering" is a thing, after all. Fact is, Google is terrified of being seen as being unfair or taking a side or ..., more than even their competitors. Their entire "our AI cannot draw white people" thing smelled more of an over-reaction to a PR threat than them trying to push a belief. Fact is, if you're seen as taking too hard a side in this election your company might be at risk after the result come down, one side or the other. Just look at how Fox is behaving since the Dominion suit. And last but not least, politics and religion are two of those subject where belief are stronger than facts and can get people very riled up very quick, so if you're an information company you want to treat it as encyclopedia factual after the fact, not as an opinion matter. I'm european, and I abhor many of the restrictions put on the current gen of LLM and image generator that come from american societal value being imposed, but on the matter of politics it doesn't matter what country in the world it's never a good idea to play that game.
- senda 3y ago
  Yeah, LLMs both in popular understanding and potential "rocket-stock" application, seem crazy over hyped. Best case it generates something somewhat accurate, which progressively decreases in probability of being accurate over time.
- aaa_aaa 3y ago
  One thing I disagree is that Google is not terrified being unfair. They are terrified because their bias would be unveiled so funny they will be shamed and ridiculed by it.
  - nolok 3y ago
    I did not say "terrified [of] being unfair" I said "Google is terrified of BEING SEEN AS being unfair or taking a side"
Jensson 3y ago
Google isn't extreme US left wing on the inside, the strongest faction is the very technical pro meritocracy side. When these two clashes you get compromises like this, the left wings wants to make the AI biased and tell you to vote left, while the meritocracy side hates that. The compromise then is just to not respond to these queries because in the end no answer is less biased than a biased answer. There isn't much else Google can do here due to their internal politics.
- input_sh 3y ago
  I'm fairly certain "extreme left" aren't the ones pushing for Biden this election cycle, they're the ones voting "uncommitted" in the primaries.
- rickdeckard 3y ago
  > the left wings wants to make the AI biased and tell you to vote left, while the [very technical] meritocracy side hates that. Any source on that to justify such a wild claim? What does "left" even mean in this context, what would be a topic that an LLM would rationally respond to in such a hard "right-wing" view that it requires someone to step in and tune it to make it respond with a "left" stance?
  - Jensson 3y ago
    > what would be a topic that an LLM would rationally respond to in such a hard "right-wing" view that it requires someone to step in and tune it to make it respond with a "left" stance? Basically anything, if you don't tune these things you get lots of porn and racist rants. You also get a lot of anti racist rants and such, but not much middle ground, and that isn't a good experience for the user.
    - rickdeckard 3y ago
      That's not an answer. Still happy to get a source on the claim that tuning an LLM in Google is a fight between a "left wing" side that wants the LLM to give a specific vote-recommendation and a "very technical side", with a concrete example.
verticalscaler 3y ago
I for one like my election information banned and/or moderated by Google, whatever is convenient for them in this moment, because they are very smart and very kind and probably know what is best.
- Jensson 3y ago
  Google search did direct you to Trumps own info pages when you searched for it instead of bombarding you with anti Trump propaganda. Google has always been relatively unbiased in these cases. But it seems like the people who do these models are a bit more extremist than the people who work on search, so the higher ups has to ban them from responding to queries like this or they would cause another uproar.
  - verticalscaler 3y ago
    > Google has always been relatively unbiased in these cases. And always will be. So we can sleep safely at night knowing a company of highly intelligent people have figured out this problem optimally and will manage it on behalf of society going forward.
  - rickdeckard 3y ago
    Or there is simply too much risk that an LLM is creating output that appears to be educated about politics and the future to a degree which might classify as election manipulation. Google has a reasonable argument that the result of a Search does not represent their opinion but is a list of what's available on the internet. But the response of an LLM can be very much interpreted as the voice of Google, because it is explicitly content THEY create...
Ekaros 3y ago
Correct me if I am wrong, but AI chatbots are mostly trained on past data? And they are not continuously being fed the newest possible information. As such with such recent events as elections it really does not make sense to use them or allow their use? As most information would be somewhat stale...
- Jensson 3y ago
  Googles gemini has no knowledge cutoff, it knows events from the past few days, I asked it about Oppenheimer the movie and it knew it won 7 oscars at the academy awards. Also asked about some famous people that died recently and it knew about those as well. So no, that is only a limitation of some chatbots, not gemini, it could absolutely be used to ask about election events.
  - pests 3y ago
    I'm sure their massive context length helps here. Could include recent events in the prompt (surfaced by Google News, Sports, Stocks, etc) until it can be properly included in training data.
    - Jensson 3y ago
      Seems to be something like that, I asked it what happened today in California and it writes stuff like this: > Other News: A new bill could see Flamin' Hot Cheetos banned from California schools [Los Angeles Times]. The LA Innocence Project is again in court pushing for DNA testing in the Scott Peterson case [Los Angeles Times]. There are reports of a worker dying after falling into a sugar vat at a factory, and parts of California are bracing for blizzard conditions [9News, The Independent]. So it tells you the headlines and gives a reference to where they got it from. Interestingly it does talk about elections as well, as long as it is a news reference: > Local Politics: Los Angeles City Councilmember Nithya Raman appears headed for a primary win based on early results [Los Angeles Times].

30 points by pg_1234 3y ago

Google is mortified by the prospect of Gemini veering off away from blanket endorsement for Establishment Democrats

This seems more like an "I'm not a threat" signal for governments than a "you shouldn't learn the bad-bad words" nanny scolding to consumers. "Preventing misinformation" and "upholding democratic values" are just the ostensible reasons for this, the deeper reason is to avoid being regulated by angry congressmen. Perhaps they have deeper reasons as well, layer upon layer of bullshit.

hnfong 3y ago
> This seems more like an "I'm not a threat" signal for governments In other places, this would be called "self-censorship".

aaa_aaa 3y ago

Perhaps they want to hide their extreme bias which would be revealed in a comical way.

politelemon 3y ago

> Sparks Debate I did not see any reference to these debates mentioned in the title. Is this just a euphemism for something the author wants to question or bring attention to without blatantly giving an opinion.

CapsAdmin 3y ago
I'm sure if pressed, the author tweeted about it from one side and got a response arguing the other side.

nolok 3y ago

It's a very useless debate. Fact is, those are LLM, not "AI" the way the general populace understand that term. Fact is, that means they just give whatever answer they can derive from their training corpus, while people think they give facts. Fact is, you do not want to be liable for your half research / half pr product being manipulated into giving wrong or distorted facts regarding one of the most controversial and pivotal election, or worse not needing manipulation just straight up hallucinating. "Prompt engineering" is a thing, after all. Fact is, Google is terrified of being seen as being unfair or taking a side or ..., more than even their competitors. Their entire "our AI cannot draw white people" thing smelled more of an over-reaction to a PR threat than them trying to push a belief. Fact is, if you're seen as taking too hard a side in this election your company might be at risk after the result come down, one side or the other. Just look at how Fox is behaving since the Dominion suit. And last but not least, politics and religion are two of those subject where belief are stronger than facts and can get people very riled up very quick, so if you're an information company you want to treat it as encyclopedia factual after the fact, not as an opinion matter. I'm european, and I abhor many of the restrictions put on the current gen of LLM and image generator that come from american societal value being imposed, but on the matter of politics it doesn't matter what country in the world it's never a good idea to play that game.

senda 3y ago
Yeah, LLMs both in popular understanding and potential "rocket-stock" application, seem crazy over hyped. Best case it generates something somewhat accurate, which progressively decreases in probability of being accurate over time.
aaa_aaa 3y ago
One thing I disagree is that Google is not terrified being unfair. They are terrified because their bias would be unveiled so funny they will be shamed and ridiculed by it.
- nolok 3y ago
  I did not say "terrified [of] being unfair" I said "Google is terrified of BEING SEEN AS being unfair or taking a side"

Jensson 3y ago

Google isn't extreme US left wing on the inside, the strongest faction is the very technical pro meritocracy side. When these two clashes you get compromises like this, the left wings wants to make the AI biased and tell you to vote left, while the meritocracy side hates that. The compromise then is just to not respond to these queries because in the end no answer is less biased than a biased answer. There isn't much else Google can do here due to their internal politics.

input_sh 3y ago
I'm fairly certain "extreme left" aren't the ones pushing for Biden this election cycle, they're the ones voting "uncommitted" in the primaries.
rickdeckard 3y ago
> the left wings wants to make the AI biased and tell you to vote left, while the [very technical] meritocracy side hates that. Any source on that to justify such a wild claim? What does "left" even mean in this context, what would be a topic that an LLM would rationally respond to in such a hard "right-wing" view that it requires someone to step in and tune it to make it respond with a "left" stance?
- Jensson 3y ago
  > what would be a topic that an LLM would rationally respond to in such a hard "right-wing" view that it requires someone to step in and tune it to make it respond with a "left" stance? Basically anything, if you don't tune these things you get lots of porn and racist rants. You also get a lot of anti racist rants and such, but not much middle ground, and that isn't a good experience for the user.
  - rickdeckard 3y ago
    That's not an answer. Still happy to get a source on the claim that tuning an LLM in Google is a fight between a "left wing" side that wants the LLM to give a specific vote-recommendation and a "very technical side", with a concrete example.

verticalscaler 3y ago

I for one like my election information banned and/or moderated by Google, whatever is convenient for them in this moment, because they are very smart and very kind and probably know what is best.

Jensson 3y ago
Google search did direct you to Trumps own info pages when you searched for it instead of bombarding you with anti Trump propaganda. Google has always been relatively unbiased in these cases. But it seems like the people who do these models are a bit more extremist than the people who work on search, so the higher ups has to ban them from responding to queries like this or they would cause another uproar.
- verticalscaler 3y ago
  > Google has always been relatively unbiased in these cases. And always will be. So we can sleep safely at night knowing a company of highly intelligent people have figured out this problem optimally and will manage it on behalf of society going forward.
- rickdeckard 3y ago
  Or there is simply too much risk that an LLM is creating output that appears to be educated about politics and the future to a degree which might classify as election manipulation. Google has a reasonable argument that the result of a Search does not represent their opinion but is a list of what's available on the internet. But the response of an LLM can be very much interpreted as the voice of Google, because it is explicitly content THEY create...

Ekaros 3y ago

Correct me if I am wrong, but AI chatbots are mostly trained on past data? And they are not continuously being fed the newest possible information. As such with such recent events as elections it really does not make sense to use them or allow their use? As most information would be somewhat stale...

Jensson 3y ago
Googles gemini has no knowledge cutoff, it knows events from the past few days, I asked it about Oppenheimer the movie and it knew it won 7 oscars at the academy awards. Also asked about some famous people that died recently and it knew about those as well. So no, that is only a limitation of some chatbots, not gemini, it could absolutely be used to ask about election events.
- pests 3y ago
  I'm sure their massive context length helps here. Could include recent events in the prompt (surfaced by Google News, Sports, Stocks, etc) until it can be properly included in training data.
  - Jensson 3y ago
    Seems to be something like that, I asked it what happened today in California and it writes stuff like this: > Other News: A new bill could see Flamin' Hot Cheetos banned from California schools [Los Angeles Times]. The LA Innocence Project is again in court pushing for DNA testing in the Scott Peterson case [Los Angeles Times]. There are reports of a worker dying after falling into a sugar vat at a factory, and parts of California are bracing for blizzard conditions [9News, The Independent]. So it tells you the headlines and gives a reference to where they got it from. Interestingly it does talk about elections as well, as long as it is a news reference: > Local Politics: Los Angeles City Councilmember Nithya Raman appears headed for a primary win based on early results [Los Angeles Times].