Hacker News Reader

Comment by alephnerd | Hacker News Reader

While everyone on HN is hyperventilating about DEI crap, the malfunction Pichai is most likely referencing is Gemini AI calling Modi a fascist. [0].

A Union Minister has already gave a veiled threat that Google broke Indian Telecom Laws [1] surrounding fake news.

It's election season in India and governments get very ban happy during this time (look at what happened to Amazon India in 2019)

Edit: Please don't make this an Indian politics flame war. The point is, a public company like Google is constrained by PR.

This is why OpenAI has been innovating so fast - they are private and don't have to answer to almost anybody

Public companies need to protect their brand, otherwise they lose out to similarly sized competitors (eg. Microsoft or Facebook)

[0] - https://www.thehindubusinessline.com/info-tech/googles-gemin...

[1] - https://twitter.com/Rajeev_GoI/status/1760910808773710038

View on HN

The link you cited, says Gemini says: 'Modi has been "accused of implementing policies some experts have characterised as fascist"'.

Is this true? Or more specifically, is this issue here that Gemini is saying something that's accurate but that the Indian government doesn't like, or that no experts have characterised Modi as a Fascist and Gemini is saying something inaccurate?

It doesn't matter.

The issue is Gemini AI spewed an answer that caused Google to enter Indian culture wars less than 30 days before national elections.

Most of the horrible crap that happens in India happens during this time of year, because Parliamentary elections are very high stakes.

Most companies want to stay out of these kinds of culture wars, and Google inadvertently entered one

How is a non generally intelligent AI like Gemini know now is a bad time to talk about Indian politics? Are models in general supposed to avoid offending anyone, anytime? Companies align the models so hard to avoid controversial opinions they lose performance. Is any kind of political question out? Is asking if Hitler was a Fascist also out?

> How is a non generally intelligent AI like Gemini know now is a bad time to talk about Indian politics

It can't and any attempt to prevent such an issue will only degrade it's capabilities.

> Are models in general supposed to avoid offending anyone, anytime? Companies align the models so hard to avoid controversial opinions they lose performance. Is any kind of political question out

I 100% agree with you.

And that's why this is such a shitshow.

Google has been so slow with rolling out LLM capabilities despite basically inventing them almost a decade ago - OpenAI doesn't have a $1.7T market cap to defend and Investor Relations to deal with.

It's only a small step from rightwing populism to fascism. Rightwing populists say some things (like "we'll let Russia invade countries that don't spend 2% of their GDP on defence"), fascists actually do it...

I guess my metric is, would you count AMLO as fascist? AMLO and Modi are functionally the same.

(Personally - I think Modi is an Authoritarian in the same mold as Indira Gandhi, Erdogan, Lula, Obran, and AMLO).

I think you're mixing up your populists here... AMLO (Andrés Manuel López Obrador) and Lula are left-wing (so you can maybe call them communists if you want to insult them, but not fascists), Orbán, Meloni, Erdogan, Putin, Modi, Trump, Bolsonaro etc. are right-wing/nationalist.

Nope. This is my point exactly.

All 3 are electoral authoritarians that came to power due to public anger at a corrupt status quo (Maluf in Brazil in the 90s-early 2000s, Coalgate/2Ggate/Jan Lokpal in India from 2011-13, and 2017 gas price hike in Mexico) and enabled by cheap populism and campaigning directly to the subaltern of their countries.

Furthermore, all 3 are products of dysfunctional democracies where criminalization of the electoral process has been the norm since the 1990s.

While electoral institutions Ann all 3 remain somewhat strong (as is reflected in the EIU's democracy index), civil society has died in India, Brazil, and Mexico.

Furthermore, all 3 leaders are notoriously anti-ideological in the sense that they will take right leaning and left leaning positions on policies at a whim depending on the electoral winds

It's only a small step from any extreme ideology to an even worse one

To be honest, whenever I read about a politician that he/she is accused of implementing policies that experts have characterized as fascist", my only thought is "Who wasn't?"

It means nothing, and I hope more and more people realize that it's all just propaganda and low-effort, mealy-mouthed slander.

There's no problem if you just cite which experts and their credentials, even better if you can add the direct quote to remove all ambiguity in the first place.

That's expensive and enters licensing issues.

Corporate License LLM do give this option, but we are also paying an arm and a leg for a Glean, Azure GPT, etc.

I would say it’s reasonably accurate to say he’s accused of it given the number of articles and related information that pop up searching for Modi and Fascism - for example, https://en.wikipedia.org/wiki/Hindutva

I don’t know enough personally to make any kind of claim beyond that though.

FTA:

>“I want to address the recent issues with problematic text and image responses in the Gemini app (formerly Bard),” Pichai said in the memo

How does your interpretation fit into the 'image responses' part? That seems to be referencing US issues.

Notice how he mentioned "problematic text" first before images.

I've worked with PR/Comms teams a lot and this kind of edit was 100% done on purpose.

Also, note that Pichai and Raghavan went to college and grade achool with a number of these mid/upper level Indian politicians.

This is another symptom of the same issue, it's not like they're disconnected. They're able to ignore the elephant in the room and act like the biased output is an affront to people of color when their models depict black Nazis after being specifically aligned to erase white people, or when they express bias in non-US politics after embedding political bias in their training, because that's what's politically correct at this time (though I'm hopeful that we're on our way to some semblance of sanity as the fallout of DEI becomes more and more ridiculous and visible). Could you imagine any big tech CEO aside from Musk criticizing the model's political bias in a US context?

Reply on news.ycombinator.com