 |
| Saint Jerome and the Angel, by Simon Vouet |
My wife went to the White Sox game Wednesday. Which caused a twinge of envy, both for her belonging to an office that socializes together, and because the Sox were playing the Guardians, the Cleveland team.
It was rainy day. And first thing in the morning, I was concerned that the 1:10 game would be rained out. So I plugged what I thought was a very simple, very direct question into Google: "Is today's white sox game cancelled?" I didn't think it would actually be cancelled so early in the morning, but it was supposed to rain all day, and I was curious to reach some site that would keep me posted. This is what AI served up:
I know this is my self-flattering bias at work. AI seems to be working just fine, helping students undercut the value of their educations, and allowing office grinds to be fired en masse, replaced by puffs of electrons up in the Cloud. And yes, I will probably still be savoring examples of AI incompetence up to the moment the robot guards herd me into the camp. But before people are going to pay money for this shit, it has to at least work, right? Forget amazing, or essential, or impressive, or even useful. This is bush league stuff.
Sometimes AI is spot on and sometimes I wonder if a bumbling idiot actually answered my inquiry. Agreed, it has to AT LEAST work!
ReplyDeleteThe same thing happened to me. I’m pretty sure it is a plot by the Trump administration because the Pope is a Sox fan.
ReplyDeleteNext time, challenge the AI when it hallucinates an answer - the response is often entertaining.
ReplyDeleteConsidering the consistent hallucinations that AI legal searches have made, causing sanctions for even some very famous well known & apparently respected law firms & that a few weeks ago a judge threw out an entire case because both sides used faulty AI that created case citations that never existed, I'm not surprised at the useless answer you got about the Sox game!
ReplyDeleteThere a couple of web sites I've commented on that have rejected the most innocent comments, because the terrible AI misread them somehow & thought they meant something else, which for the life of me I couldn't figure out what?
Yesterday, someone called Mister S an "ignorant slut"...a reference to Jane Curtin, a member of the original SNL cast in the 70s. It's the AI bots that are the ignorant sluts. They have no knowledge of how people actually speak. They do not understand, slang, idioms, or colloquial phrases.
DeleteWas once flagged and booted because the mods policed their page via AI chatbots. An old expression for a substance addiction (in this case, it was heroin) was "a monkey on my back." The bot thought the "monkey" was a person of color and that I was a hooded and robed racist! You can't make this shit up. You also can't use the m-word online anymore.
AI is invasive and corrupts true history groups when stolen images and AI-generated, florid (and often inaccurate) captions are reposted from sites by clickbait-seeking clones, which are run from other continents by people who have never been to the cities and states they portray. Admins need to be removing such reposts from those pages and sites.
If you like those scamming and data-mining places, great...join them and be scammed, hacked, and misinformed about the past. But don't engage, and encourage, or respond to these charlatans,. Help make American historical pages great again!
There's a distinction between choosing to use AI and using it unknowingly. If you opt to use AI, the onus is on you to verify and fact-check. If you use it unknowingly, you have been duped! I don't think it's unreasonable to require a disclaimer on AI-generated material, or to allow someone to block AI use in their browsers. I still marvel at the enormous gaffe made by the SunTimes* last summer when they published a list of the best summer reads and a large number of the books were hallucinations. There was a new book by Andy Weir on the list that looked good. I tried to purchase it as a gift and was duped!
ReplyDelete* The book list was generated by a subcontractor who used AI to create a supplement for the paper.
I've always said, any AI material must be clearly and easily identified, under a rule/law with penalty tbd.
DeleteIf you look at the most important metric regarding AI you will find that it is functioning even better than anyone could have expected.
ReplyDeleteIt is reducing headcount, frustrating people trying to fix issues with major companies, and making a select few individuals very very rich.
I have found that AI can help edit my poorly written slop, but i have to go back and reread it to make sure it hasn't removed too much. Searches still require double and triple checking and I tend to use it more like google of old to find the place(s) that has the information i need. AI seems to be "helpful" to a lot of people I know. Many of them can now more easily do the job of three than they could before. It is the perfect late stage capitalism tool; something you don't have to treat well because it isn't real, you don't have to pay because it isn't a person, and a relatively fixed cost compared with other "workers."
AI will slowly put millions out of jobs and the top 1% or the 1% won't even realize.
Let them eat digital cake.
AI is also setting massive amounts of money afire, it's likely the single greatest losing proposition the business community has embraced.
DeleteWriters Cory Doctorow (check out pluralistic.net) and Ed Zitron, among others, who follow it much more closely than I, or most, do, both dig into the hype.
One great illustration is Doctorow's comment that if the boss doesn't show up, work continues as normal, while if the workers don't, it all stops.
They illustrate it better than I can summarize it here.
Even the one percenters will suffer if a million customers disappear.
Deletetate
Alan and Tate, you are both right. Though I do wonder if they will be around to see those losses. Look at Elon Musk; by nearly every metric the man has become a massive failure. And yet, he is -- or was -- the first trillionaire. The economic gods seem to smile so brightly on him that even his over extended, operating in the red, twenty years behind schedule companies are deemed to be worth 100x what they should be.
DeleteMusk can burn $50 million dollars a day and wouldn't run out of money for 50 years... that's nearly unfathomable.
Since AI can't be embarrassed when it's wrong, shouldn't we be embarrassed to use it?
ReplyDeleteYou are quite right, NS.
ReplyDeleteNot to stir up trouble, but referencing a past post, the Niner's stadium has some familiar signage, with pretty good author commentary: "But most importantly, at every stadium, like most sports reporters, I like to sample a little urinal water. The Niners don't allow that. Libs trying to take away guns, freedom, and urinal water, smh."
ReplyDeletehttps://www.phillyvoice.com/official-nfl-press-box-food-spread-rankings-cowboys-jaguars/
Yes, bots like ChatGPT do make mistakes, and hopefully they’re obvious. I wonder if Mr. S. replied that there was indeed a game today, go look again. He would probably have gotten an apology and a much more accurate response.
ReplyDeleteThis assumption that AI is somehow universally stupid is already outdated maybe the AI that they make available to you for free when you use a search engine but if you have a subscription and I don't mean the cheap one better regular person might buy but a corporate institutional AI program that s*** is a hell of a lot smarter than anybody you've ever known
ReplyDeleteTom