General Discussion
Related: Editorials & Other Articles, Issue Forums, Alliance Forums, Region ForumsThis Is What It Looks Like When AI Eats the World - Charlie Warzel in The Atlantic
Gift link from his tweet about it below (and see this earlier thread about OpenAI's deal with the Atlantic - https://www.democraticunderground.com/100218988065 ).
https://www.theatlantic.com/technology/archive/2024/06/ai-eats-the-world/678627/?gift=bQgJMMVzeo8RHHcE1_KM0Z6IC-mVnbZ1MIab37oJyhY&utm_source=copy-link&utm_medium=social&utm_campaign=share
Technology companies, in other words, are racing to capture money and market share before their competitors do and making unforced errors as a result. But though tech corporations may have built the hype train, others are happy to ride it. Leaders in all industries, terrified of missing out on the next big thing, are signing checks and inking deals, perhaps not knowing what precisely it is theyre getting into or if they are unwittingly helping the companies who will ultimately destroy them. The Washington Posts chief technology officer, Vineet Khosla, has reportedly told staff that the company intends to have A.I. everywhere inside the newsroom, even if its value to journalism remains, in my eyes, unproven and ornamental. We are watching as the plane is haphazardly assembled in midair.
-snip-
Theres an element of these media partnerships that feels like a shakedown. Tech companies have trained their large language models with impunity, claiming that harvesting the internets content to develop their programs is fair use. This is the logical end point of Silicon Valleys classic Ask for forgiveness, not for permission growth strategy. The cynical way to read these partnerships is that media companies have two choices: Take the money offered, or accept OpenAI scraping their data anyway. These conditions resemble a hostage negotiation more than they do a mutually agreeable business partnershipan observation that media executives are making in private to one another, and occasionally in public, too.
-snip-
Nobody knows whats coming next. Generative-AI companies have built tools that, although popular and nominally useful in boosting productivity, are but a dim shadow of the ultimate goal of constructing a human-level intelligence. And yet they are exceedingly well funded, aggressive, and capable of leveraging a breathless hype cycle to amass power and charge head-on into any industry they please with the express purpose of making themselves central players. Will the technological gains of this moment be worth the disruption, or will the hype slowly peter out, leaving the internet even more broken than it is now? After roughly two years of the most recent wave of AI hype, all that is clear is that these companies do not need to build Skynet to be destructive.
AI is eating the world is meant, by the technologys champions, as a triumphant, exciting phrase. But that is not the only way to interpret it. One can read it menacingly, as a battle cry of rapid, forceful colonization. Lately, Ive been hearing it with a tone of resignation, the kind that accompanies shrugged shoulders and forced hands. Left unsaid is what happens to the raw materialthe foodafter its consumed and digested, its nutrients extracted. We dont say it aloud, but we know what it becomes.
Much more at the link. I hope you'll read all of it.
OAITW r.2.0
(32,163 posts)Stay tuned.
Celerity
(54,427 posts)usonian
(25,379 posts)No, playing Transylvanian Lullaby won't tame it.
https://www.wired.com/story/google-and-microsofts-chatbots-refuse-election-questions/
https://archive.ph/PnG8Q#selection-595.0-601.200
Googles and Microsofts AI Chatbots Refuse to Say Who Won the 2020 US Election
With just six months to go before the US presidential election, Gemini and Copilot chatbots are incapable of saying that Joe Biden won in 2020, and wont return results on any election anywhere, ever.
https://www.xda-developers.com/microsoft-recall-pr-nightmare/
Microsoft is revamping how Recall works amid its PR nightmare
Anyone could have predicted this
Microsoft is a company that's actually pretty good at security, so it's surprising that the company was ready to send Recall out the door with issues that were obviously going to upset users. Moreover, the service was only announced a couple of weeks ago, and the major part of the backlash is just days old. It didn't take the firm long to come up with a fix.
So, it would seem that in the entire time that Copilot+ was in development, this never occurred to the Windows team. One would have to ask how that's possible.
Interestiingly, this hasn't changed the launch date of Copilot+ PCs, which are still going to hit shelves on June 18.