I've just tried the new #thunderbird Beta and... wow, I'm completely blown away. It looks amazing, seems to have threading, and is blisteringly fast on my terribly overloaded inboxes. Bravo to the @thunderbird team!
The #AI Trilemma: or why AI won't be as revolutionary as we expect.
We are in the midst of an AI revolution... or are we? As business leaders around the world scramble to integrate AI into their products and business practices, the cracks are beginning to emerge. We've seen GPT4 fail at basic reasoning, Google's AI search telling people to jump from a bridge, Microsoft's copilot inventing quotes from Vladimir Putin, and recidivism algorithms that continue to send the wrong people to prison.
That's not to say that there aren't successful applications of AI: there are in certain niche areas, but I don't think AI will be as revolutionary in our everyday lives as expected for one key reason, which I call the AI Trilemma.
Essentially, when developing an AI system you can pick any 2 of three qualities: Economic Viability, Safety/Reliability, and Effectiveness.
In order to build an AI that is both Effective and safe/reliable, the types and quantities of data needed are not available or would require prohibitive investments to achieve.
In order to build an AI that effective and economically viable, you have to forgo safety, or accept that you are using potentially incomplete or biased data, which could result in harmful output.
In order to build an AI that is Economically Viable and Safe/Reliable, you have to build in so many limitations that you may as well not use AI at all and stick to traditional software, as it will be more effective at a lower cost.
I might be completely wrong about this, but as it stands the data to build safe and reliable AI doesn't exist: scraping reddit to train an AI won't produce quality results, nore has scraping the wider internet. Building a bullshit generator whose usefulness diminishes by the day and marketing it as general purpose AI isn't a sign we're heading for a bright future: it's a sign we are in the midst of an AI bubble, and it's only a matter of time before it pops...
@Di4na I think it very much depends on what the company is trying to build. For instance, Firefox's new image recognition can work as intended for the most part, but the false claims that we are going to build some sort of geature-filled universal AI is BS
Translation: "We share the horror of today's assassination of the Prime Minister [Fico of Slovakia]. We fear a further escalation of tensions in society. We therefore warn against the dissemination of false information about the shooter. We unequivocally deny that he is a member of our movement. Nor is there any other connection whatsoever between him and our party or our members. We strongly condemn his heinous act."
Featured speaker this time was Euro Member of Parliament @kimvsparrentak who passionately spoke on the need for a more free Internet detached from #bigTech and more focussed on #Democracy and real #Social interaction.
When confronted with her presence on X and Insta, and the lack of active presence here on the #Fediverse we promised to help her.
Please help us get a decent follower-base here and follow her account right now, before she even gets a chance to get more active. A follower base of more than a 1000 people will surely help her convince her party members to spend time and money on a real presence here.
The Commission is approved (and can be rejected) by the Parliament, just like the government can he approved or rejected by parliament in the Netherlands