The Daily Beast reports that “The Los Angeles Times removed its new AI-powered “insights” feature from a column after the tool tried to defend the Ku Klux Klan.”
The Daily Beast reports that “The Los Angeles Times removed its new AI-powered “insights” feature from a column after the tool tried to defend the Ku Klux Klan.”
8 comments:
Despite the the KKK history, AI considered them not explicitly "hate-driven movement".
Today their sentiments live strong in M.A.G.A and the republicans despite the lack of violence.
In other opinionated news,republicans are destroying everything. I can't think of one thing the trump republicans have done right except maybe stronger border control. I'm a democrat.
Don’t you see, though? It’s just not okay having a bunch of unelected people administering the law. But it’s totally fine to let a bunch of computers administer our laws. And all the better that we don’t even really understand how the technology works!
AI is not doing anything different than most of corporate media. They are parroting the talking points of this administration. Has anyone seen the briefings from the White House? The MAGA reporters involved but yet AP has been banned? How about the recent debacle in the White House where MJT’s boyfriend was asking a question to Zelensky in regards to his attire or a Russian news agency given access to cover the historic “mineral” deal? These corporations are scared of being sued and they are being nice little robots.
The MAGA folks would see this as a feature, not a bug.
AI functionality consists of a lot more than generating opinions that can contain hallucinations or inappropriate content. It should be easy to identify what functions are potentially harmful or unreliable and restrict those uses. For instance, AI predicting/determining outcomes for disability decisions should be taboo - not even pursued, let alone implemented. Or generating language that is shared with the public without humans in the loop - it's not ready for that, as the example in the LA Times illustrates. But it is a fantasy to think that AI can't help (revolutionize) many, many things that are done internally as part of our workloads. It will soon be able to generate a summary of medical evidence much better than anything we can do "by hand." Most importantly, it is getting better all the time, and what might be expensive state-of-the-art capabilities now will be commonplace soon enough. It will filter down to the rest of us. Also, critically, it should be obvious enough by now that the current administration is hellbent on bringing a Silicon Valley attitude and methods to our work. All of those factors mean we can't count on shielding ourselves from change the way we have in the past. It's here now.
Please don't cite IMAGEN or whatever else SSA has done so far or funny results you get on your own version of chatGPT as a rebuttal, because that is not a fair representation of what current AI is capable of. It is doing the work of high-level medical researchers now - you think it can't handle a straightforward summary of an SSA file for our own internal purposes, so that ALJs can get right to the heart of their job without exhausting themselves reviewing 1500 pages of evidence? Not to decide the outcome, but to report its contents? You think it can master the intricacies of protein folding but can't wrap its head around SSA eligibility requirements?
I am NOT a DOGE apologist, and I AM fond of the rule of law, and I believe in SSA's mission and carrying it out as a govt agency - not privatized/for profit. But AI will be a fact of life from here on out in the same way electricity is - it's way too late to turn back. And in any event, it's out of our hands as mere govt employees, especially in the current environment. Even at this stage of development, state of the art AI can write as well as I can for straightforward purposes, and it is getting better all the time. I wouldn't trust sharing its output with the public without an editor, but it's time to open our eyes and figure out how we can adapt it for our mission most effectively. Get on board now or risk being left behind. Just focus on our mission - processing claims as fast and as accurately as possible. If there is something that helps do that, embrace it. Adapt to the new world now - the old one is on fire. Don't concede the government's functions to the profiteers by holding on to the old ways. That's not a fair fight. We need the efficiency of private industry, including AI, with the public service mentality and motivations of good, compassionate government.
Yes let AI handle 800 calls. And wait for the complaints to flood into Congressional offices...
My favorite fact about chatgpt is that if you ask it how many ‘b’ are in “strawberries” is it will insist there are 2 of the letter b. Sometimes it might tell you there are 3. But sure, let’s give it all right on over to “AI”…
Can AI be any worse that your average VE? Pan Greaser? Garment Sorter?
Post a Comment