News for hackers. Stuff that matters.
Re:Gemini: Google's most capable AI model yet by brainwad on Wednesday December 06, 2023 @04:17PM (#38549584)
Launching anything as a big tech company in Europe is an absolute nightmare. Between GDPR, DSA, DMA and in Google's case, several EC remedies, it takes months to years to get anything launched.
Re:Gemini: Google's most capable AI model yet by imiric on Wednesday December 06, 2023 @07:06PM (#38551192)
It's only a nightmare if you are an adtech company whose revenue relies on tracking users, and have a history of violating their privacy.
Re:Gemini: Google's most capable AI model yet by taneq on Wednesday December 06, 2023 @07:41PM (#38551456)
While that's usually my line too, in this case it's also a nightmare if you're selling access to an AI and it unexpectedly starts barfing up EU citizens' private data verbatim that 'somehow' ended up in the training set.
Re:Gemini: Google's most capable AI model yet by pb7 on Wednesday December 06, 2023 @08:51PM (#38551879)
It's a nightmare for anyone that isn't a tiny startup that flies under the radar. See: shocking lack of global-scale innovative companies made in Europe.
Re:Gemini: Google's most capable AI model yet by usef- on Wednesday December 06, 2023 @09:55PM (#38552264)
I hope you're right. But do you have an alternative explanation for why there's seemingly far fewer companies coming from the EU?
Re:Gemini: Google's most capable AI model yet by kolinko on Thursday December 07, 2023 @03:59AM (#38554369)
Language barriers and local markets. 95% startups here build things for local markets and fail to scale to the global ones. In the meantime US startups scale to ~1bn people almost instantly.
Re:Gemini: Google's most capable AI model yet by Postosuchus on Wednesday December 06, 2023 @11:06PM (#38552702)
Re:Gemini: Google's most capable AI model yet by krzyk on Wednesday December 06, 2023 @04:25PM (#38549684)
OpenAI somehow managed to do a release worldwide.
Re:Gemini: Google's most capable AI model yet by nmfisher on Wednesday December 06, 2023 @07:35PM (#38551404)
OpenAI isn’t a monopolistic behemoth like Google with a cash cow to protect. Google attracts far more attention and has a lot more to lose.
Re:Gemini: Google's most capable AI model yet by cyanydeez on Wednesday December 06, 2023 @07:15PM (#38551270)
by not having lawyers and believing they did nothing wrong.

you typically see brazen behavior from ignorance.

Re:Gemini: Google's most capable AI model yet by tazjin on Wednesday December 06, 2023 @04:40PM (#38549893)
For some values of "world".
Re:Gemini: Google's most capable AI model yet by buzzert on Wednesday December 06, 2023 @04:53PM (#38550054)
They got banned in Italy pretty quickly.
Re:Gemini: Google's most capable AI model yet by ithkuil on Wednesday December 06, 2023 @06:11PM (#38550759)
But the ban didn't last long (I'm using openAI from Italy)
Re:Gemini: Google's most capable AI model yet by dataking on Wednesday December 06, 2023 @09:27PM (#38552092)
A startup doesn't have the same target of its back as a large publicly traded corporation. It also has a different culture and is expected to take risks that non-startups aren't. In other words, not an apples-to-apples comparison.

Edit: grammar.

Re:Gemini: Google's most capable AI model yet by sho_hn on Thursday December 07, 2023 @04:38AM (#38554584)
Far more likely answer: OpenAI isn't interested in withholding a product to play political cards.
Re:Gemini: Google's most capable AI model yet by krzyk on Friday December 08, 2023 @11:26AM (#38570809)
I think this is more related to american-centrism of Google - it is in their other products - e.g. Pixel phones restrict certain features to US.

So was Stadia - restricted to US at the beginning.

Also their are very slow at adding non-English languages to their offering.

Re:Gemini: Google's most capable AI model yet by crucialfelix on Thursday December 07, 2023 @01:22AM (#38553455)
But Anthropic still isn't available here. We will see what happens to Mistral (paris, $2B valuation !)
Re:Gemini: Google's most capable AI model yet by dash2 on Wednesday December 06, 2023 @04:29PM (#38549732)
Two sad things:

1. This stuff is available in like Angola and Thailand but not in Germany or France. Oh how the European giant has fallen.

2. ... but it's also not available in the UK. So the long shadow of EU nonsense affects us too :-(

Re:Gemini: Google's most capable AI model yet by martimarkov on Wednesday December 06, 2023 @08:31PM (#38551777)
Yeah having privacy protections for everyone really hurts us… Someone releases something shiny so we should just allow them to harvest personal data and manipulate markets so we get access to it… not really a society I think most giants want to live in. I prefer waiting a bit but knowing that Google needs to play on a bit more even field over here. Plus none of the GDPR or DMA are that bad. Just make sure you comply and get it over with. It’s not that hard to build a privacy centric product that doesn’t steal my data.

On 2 yeah it does. Seems like the UK keeps falling behind on everything now that it lives in the shadow of the continent and can’t seem to create any value and nobody cares about that market. So much for the MaSiVe TraDe DeALZ we were getting unlocked…

Re:Gemini: Google's most capable AI model yet by thefourthchime on Wednesday December 06, 2023 @11:02PM (#38552685)
Sorry, I had to click on a GDPR before I could read this so I clicked away.
Re:Gemini: Google's most capable AI model yet by pb7 on Thursday December 07, 2023 @12:14AM (#38553096)
> Just make sure you comply and get it over with. It’s not that hard to build a privacy centric product that doesn’t steal my data.

Eh, given that Europe struggles to build anything tech related, I'm going to say it's pretty hard. Far easier to make overpriced luxury handbags and the like.

Re:Gemini: Google's most capable AI model yet by sho_hn on Thursday December 07, 2023 @04:34AM (#38554567)
Aside from, you know, most likely making the machines required to build the computer you're spewing the xenophobia on, and countless other examples that aren't making tech as simple as websites.

The anti-European gloating on HN is getting tiresome, and imho is a pretty big blind spot of HN moderation.

Re:Gemini: Google's most capable AI model yet by dash2 on Thursday December 07, 2023 @04:59AM (#38554700)
I think it's very exaggerated to call this xenophobia and suggest it needs to be moderated. OP said Europe struggles to make tech, and focuses more on luxury. This isn't xenophobic - it's true.
Re:Gemini: Google's most capable AI model yet by sho_hn on Thursday December 07, 2023 @05:07AM (#38554748)
> I think it's very exaggerated to call this xenophobia

I'm also not quite sure it's xenophobia, but it's something other than calm and rational. Whenever Europe comes up in a HN thread, there's a sudden glut of snide comments and gloating, as if people are desperate for themselves and others to believe their layoffs-stricken industry is amazing nonetheless. Maybe to distract themselves from the mandatory RTO bad news?

Rather than calling it xenophobia, let me call it adversarial, and it's always the same side initiating it. Very, very tiresome.

I'm also willing to bet a very high % of the same demo would readily call HN an explicitly American forum and that inclusiveness or being welcoming to others shouldn't be a priority of the site, which I personally find just sort of jarring anywhere on the web, since https doesn't know borders and I grew up in the very overtly international FOSS community as an engineer. You know, where we make that Linux thing you probably have ten copies of on devices around you right now that started in Finland.

> OP said Europe struggles to make tech, and focuses more on luxury.

Only for a very narrow definition of "tech" that doesn't include the vast majority of engineering disciplines, as well as basic research and the education that enables the "tech".

I will take, for example, BioNTech over most startups posted on HN, and I'd rather we build another ITER than another Twitter clone.

Your bio says you're an R programmer. Are you aware the R foundation is a European (Austrian) org, like most organized open source? Seems those Europeans have quite a bit of time for tech after all.

Re:Gemini: Google's most capable AI model yet by red75prime on Thursday December 07, 2023 @05:14AM (#38554823)
Europe needs a new law to regulate moderation, obviously.
Re:Gemini: Google's most capable AI model yet by behnamoh on Wednesday December 06, 2023 @08:44PM (#38551839)
> It’s not that hard to build a privacy centric product that doesn’t steal my data.

Dude, we're talking about Google here...

Re:Gemini: Google's most capable AI model yet by cyanydeez on Wednesday December 06, 2023 @07:14PM (#38551261)
"raping and pillaging intellectual and privacy domains are harder where protected"
Re:Gemini: Google's most capable AI model yet by thefourthchime on Wednesday December 06, 2023 @11:00PM (#38552677)
That and the labor laws and very strict compared to the US. We have an office in Poland due to regulations most "employees" want to be contractors, this is due to a tax advantage / national healthcare I think. That said the offices in the US don't say it out loud, but in the EU seem to take a week off every other month for something.

At the end of the day, the employees have a much cushier life-work balance. You can argue (rightfully) that that's better for the people and society, but it also means it's harder for companies to succeed.

Re:Gemini: Google's most capable AI model yet by kolinko on Thursday December 07, 2023 @04:03AM (#38554396)
Being contractors is Poland - that’s because of the taxes exactly, not regulations per se.

Contractors get taxed 19% flat rate and a small% for health insurance (even less in IT - 9%?). Whereas full time workers get taxed similarly to people in the west.

Re:Gemini: Google's most capable AI model yet by eastern on Thursday December 07, 2023 @06:26AM (#38555241)
Launching a small company is an even bigger nightmare, and that's actually the bigger problem.

The legal cost of dealing with a few _mistaken_ (or fake) GDPR complaints can wipe you out.

The bigger company will have inhouse or retainered lawyers who'll deal with it.

Almost all regulation acts as a barrier which protects bigger companies who can pay lawyer fees without blinking.

It's amazing how much of the HN crowd sides with the bureaucracies which are basically pals of the guys with deep pockets.

Re:Gemini: Google's most capable AI model yet by sensanaty on Thursday December 07, 2023 @07:52AM (#38555774)
> The legal cost of dealing with a few _mistaken_ (or fake) GDPR complaints can wipe you out.

No, they can't. It's not an automated system that automagically fines companies if they get flooded by fake emails or whatever, they're pretty reasonable most of the time and you get given plenty of chances to work with regulators before they decide to fine you even a single euro (assuming you're guilty in the first place). Even if you get fined, they're usually scaled to the severity of the offense as well as the company's size.

Plus the solution is super simple, just don't invasively track your users without consent! I love that I can use the GDPR to tell my manager to fuck off when he talks about using some invasive tracking bullshit on our users, I'm glad it's there.

I'm not sure if this[0] is the most up-to-date list (I've seen a number of these lists), but take a look yourself. Most of these fines are tiny, certainly not earth shattering for any company of any size with any stability.

And if your business can't survive the financial burden of complying with GDPR, then good. There's no reason for a small business to even be violating it in the first place, since we've had about a decade of forewarning at this point regarding these privacy laws.

===

[0] https://gdpr-fines.inplp.com/list/