Elon Musk: Always A Douche

[crazy canuck]

They just can't fathom, that critical reporting of all politicians is sort of what the CBC does.  Its just that the right thinks that means the CBC is taking directions from the left leaning political masters.  But for some reason the PPs of the world, and apparently you, ignore all the critical reporting the CBC does of the NDP and Liberals, not to mention how the CBC's reporting of the Greens debacle was withering.

There is much less criticizing of the Liberal party or the NDP than of the Conservatives or the Bloc at the Federal level.

They even got blamed for their lack of impartiality at the last electoral debate but they refused to apologize for it.

I am not aware of your second point, do you have a link?

I understand that your first point is an opinion widely held on the right.  But do you have anything to substantiate it?

[OttoVonBismarck]

Not to interrupt the Canadian political thread (and I do enjoy the political ramblings of our client state to the North)  ; but it looks like due to the fact Musk has recently massively increased charges to use the Twitter API a number of enterprise partners are deactivating links to the service. My understanding is most, maybe all? of the Twitter API access was previously free.

Musk saw this as untapped revenue.

That isn't entirely unreasonable--services like this are often straddling a fine line with these APIs. On the one hand, free API access gets your service linked to lots of other things, which can help it grow and maintain users. On the other hand, API use often leads to users who are gleaning stuff off the service but not really using the service itself (and importantly--this means they aren't seeing the service's impression ads.)

The line a lot of similar companies seem to set is that limited / small scale API use is often allowed free of charge, enterprise level use is often charged some fee. The amount of fees is going to vary a lot based on a number of factors, but it seems like as of right now, the enterprise API fees Musk has set for Twitter are high enough a lot of pretty big companies are just deciding to end integrating with Twitter over it.

I've heard stories of some large enterprises looking at $50k/mo API fees or so. Now, there's plenty of large organizations that pay big subscription fees of that size, and some much more--but when you go from free to imposing a fee, you're forcing a decision point for the enterprise. Previously this cost nothing, now it costs say, $600k/yr, that means at the last someone with budgetary approval has to agree it is worth continuing to use it. Considering Twitter has often been seen by advertisers and basically "everyone" as one of the "least valuable" social media platforms, it is no surprise that forcing this decision point is leading to some large enterprises to just wash their hands of Twitter.

Microsoft in one of its sales / marketing platforms recently did just that, they told customers they would no longer support the Twitter integration after a certain date.

In response, Musk has Tweeted that Microsoft was using its data "illegally" and now it is "lawsuit time."

I am genuinely confused if Musk thinks you can just sue companies for "deciding not to buy your product or service."

It reminds me that basically all of Musk's commercial success has not involved having to be friendly with enterprise customers. SpaceX customers are heavily tilted towards governments, Tesla is largely selling direct to consumers. Enterprise software sales is a pretty complex field and not easily navigated--there is a reason salesmen in that space can easily make $300-400k+ a year, and threatening to sue potential customers is probably not on any kind of shortlist of good ideas for how to make those kinds of deals.

[Jacob]

I wonder if Musk thinks he has a case or whether he thinks Microsoft is an easy target for a legal browbeating? Or, I guess, the third option is he just tweeted random shit without thinking it through.

[The Minsky Moment]

The theory of the hypothetical lawsuit is that ChatGPT allegedly used Twitter data to train its model.  not entirely clear how they get to MSFT, which is just an investor, but it may be based on the integration with Bing. 

[Jacob]

Ah that makes some sense. There are probably some big legal pay days coming (for lawyers at least) regarding AI training on data sets.

If ChatGPT did use Twitter data for training, how much would the profitability of ChatGPT likely weigh on the outcome (as in, if ChatGPT hasn't made much money yet how much can Twitter get out of a lawsuit)? And, again, if ChatGPT did in fact use Twitter data to what degree could they say "okay, nevermind, discard ChatGPT, we're going to make a new ChatGPT using different data sets to lessen the repercussions?

[Sheilbh]

The theory of the hypothetical lawsuit is that ChatGPT allegedly used Twitter data to train its model.  not entirely clear how they get to MSFT, which is just an investor, but it may be based on the integration with Bing. 

Working for a media company, I can only laugh and cry at a social media company wanting to be more in control of and get paid for other companies extracting commercial benefits out of your content

[Jacob]

"It's ours! We stole it fair and square!"

[Hamilcar]

The real nightmare is that Europeans seem to have a right to force OpenAI et al. to remove their personal data from training under GDPR.

[OttoVonBismarck]

Mind you also--the marketing software the Microsoft decided to cut Twitter API integration with AFAIK had nothing to do with ChatGPT. That software essentially integrated with Twitter's direct message API, to handle corporate Twitter accounts that receive DMs from customers etc. This is basically just Musk threatening to sue over an unrelated grievance because he is mad Microsoft isn't giving Twitter money.

Microsoft != ChatGPT, though--it is just an investor in the firm behind ChatGPT. Bing does use ChatGPT so maybe you could make some claim there--it sounds fairly specious though, AFAIK ChatGPT trained on information that was publicly shared on the internet.

I don't know how much Microsoft would be considered a "soft target" for a speculative grievance lawsuit--they appear to regularly litigate complex, sometimes decade-long cases dealing with vague areas of the law as it relates to software companies.

[Jacob]

The real nightmare is that Europeans seem to have a right to force OpenAI et al. to remove their personal data from training under GDPR.

By nightmare you mean "this is super inconvenient for people training AI", right?

[Hamilcar]

The real nightmare is that Europeans seem to have a right to force OpenAI et al. to remove their personal data from training under GDPR.

By nightmare you mean "this is super inconvenient for people training AI", right?

Also for existing AI. A deletion request for a person's data from GPT4.... No idea how to do that.

[The Minsky Moment]

Ah that makes some sense. There are probably some big legal pay days coming (for lawyers at least) regarding AI training on data sets.

If ChatGPT did use Twitter data for training, how much would the profitability of ChatGPT likely weigh on the outcome (as in, if ChatGPT hasn't made much money yet how much can Twitter get out of a lawsuit)? And, again, if ChatGPT did in fact use Twitter data to what degree could they say "okay, nevermind, discard ChatGPT, we're going to make a new ChatGPT using different data sets to lessen the repercussions?

It's hard to speculate about what the precise legal basis of this lawsuit would be.  In theory you could postulate a copyright violation, and US copyright laws permit recovery of statutory damages per incident of infringement.  So, e.g., if ChatGPT used millions of records without permission, the damages exposures could be in the billions or even trillions of dollars.

Of course, the obvious problem is that twitter can't claim copyrights in the content of tweets made by others on the platform.  So I'm not exactly sure what they are up to.

[Sheilbh]

The real nightmare is that Europeans seem to have a right to force OpenAI et al. to remove their personal data from training under GDPR.

By nightmare you mean "this is super inconvenient for people training AI", right?

It depends. The right to be forgotten isn't absolute.

Also just worth flagging that GDPR rights aren't just for Europeans. It's not like, say, California privacy law which is about the rights of Californian residents. Data subjects can come from anywhere - if you sign up for, say, Spotify, because they are subject to GDPR all individuals whose data is used by them have the same rights.

I'm not 100% sure that every AI based in the US would necessarily be subject to GDPR - or perhaps if the tests of who's caught would necessarily make sense with AI. The Italian regulator had a few points on their decision for OpenAI to stop in Italy (reports are talks are going well and they'll probably back in the next few weeks) - some are fairly fundamental but the headline seemed to be the lack of an age gate.

[celedhring]

So, you made me ego-search myself on ChatGPT and I got:

I'm sorry, but I do not have any information about an individual named XXXXXXXXXXXX. It's possible that he is a private individual with no significant public presence or contributions, or that he may be a relatively unknown person in history. If you have any additional context or details about this person, I might be able to help you further.

Which is interesting, since there's stuff about me easily found online (imdb, linkedin, etc...).

[HVC]

So, you made me ego-search myself on ChatGPT and I got:

I'm sorry, but I do not have any information about an individual named XXXXXXXXXXXX. It's possible that he is a private individual with no significant public presence or contributions, or that he may be a relatively unknown person in history. If you have any additional context or details about this person, I might be able to help you further.

Which is interesting, since there's stuff about me easily found online (imdb, linkedin, etc...).

Maybe it was just being sassy and gave you the Mariah Carey treatment?