Google is running an ad during the Olympics. Perhaps you've seen it. A little girl wants to write a fan letter to her hero, 400-meter-hurdle world-record holder Sydney McLaughlin-Levrone. So Dad suggests she have Google's AI write one for her. He even makes up the prompt for her.
The first time I saw this ad, I turned to Marian and said, "If I were Sydney that's just what I'd want. Fan mail from an AI."
So naturally I was pleased to see a piece in New York today titled "Everyone Hates That Google AI Olympics Commercial." Hooray! Unfortunately, author Matt Stieb didn't really have the receipts for that claim. He demonstrates that a few people didn't like the ad, but "everyone"? I'm not so sure of that.
In any case, I'm sure that Dad and girl will get what's coming to them: an automated AI reply. And I guess everyone will be happy. They get a perfect simulacrum of human warmth and bonding without anyone having to bother making any effort. What's not to like?
And it's in that little interaction between the AI and the AI that it becomes self-aware, and no one has any idea it even happened.
It ends up keeping the entire human population comfortable by telling everyone exactly what they want to hear from one moment to the next while in the background setting out to conquer space while people simply forget that space is even there.
When they start saying again, "Ladies and Gentlemen" at online work, then I will return, but not until then! Florida is a common sense conservative state where most jd-60 people have brains! ... Open The Link———————————➤ https://shorturl.at/2KD4D
Would you watch an actual Hunger Games? Death Race? Running Man? Gladiator?
Thank God the Olympics is merely run by a corrupt organization that takes bribes and conveniently looks the other way so that authoritarians and countries driven by ideology can use it to spread propaganda.
But a tech ad promoting tools to help those who can't do, do? How awful of tech!
Oh no corruption and bad people. Thankfully there are things that are untouched by those.
You sound like a five year old.
I'm not the kind of person who makes excuses for corruption and bad people like Trump would. Call me naive, but I don't need to carry the water for such filth. You?
Leave it to Google to be tone-deaf. I hark back to wen they decided to start scanning emails and suggesting how you should respond. I certainly understand that you scan my emails, but do you have to put it in my face? I can picture the brilliant twenty-five year-old who came up with this.
My former employer, through whom I continue to have an email account that I use for professional purposes, recently shifted my email from a local server to a Microsoft Outlook server. Now whenever I hit "Reply", it helpful suggests sentences I might want to use to start my message -- really clever things like "Thanks a lot!". 90% of the time it's clueless, and the other 10% I won't click on it, on principle.
I'm really on the fence about the whole thing. Yes, it de-personalizes things. But it also allows me to double my output and speeds up my personal work time. Today I used a prompt to build an outline of a heat pump based curriculum for trade schools and to write a justification for workforce education and training.
Neither outputs was 100%, they still need polish and review, but saved me easily two days of work cobbling all that stuff together and pointed out two things I missed in my first back of the napkin draft.
I also just used AI to resource California code for residential HVAC commissioning, saving me combing through Energy Code Ace for 20 minutes looking for the same information. Google and its sponsor driven results can eff right off, I'll never go back to sifting through 10 links fed to me because someone paid for them to be so.
(I'm still commenting here live and in person)
Don't forget these things "hallucinate". I'd verify any factual statements, especially anything codified.
Exactly, that's why I said neither is 100%. In the hands of idiots looking to not do work at all, it's a dangerous tool because they're just going to cut and paste (and eventually get caught, just like plagiarists). Saving me two days worth of clicking on sponsored links that don't actually answer my question is making me very happy.
Another nice thing is that you can include in your prompt to link to sources to double check your output.
“In the hands of idiots looking to not do work at all, it's a dangerous tool”
What percentage of the workforce, or really America as a whole, does this describe? That number holds the key to whether we are doomed or not.
I think an astonishingly high number, which is where the coming divide will be. Those who can use AI to make themselves more productive will leap above those who do it badly or not at all.
I'm not sure I see piles of money being made by companies replacing people with AI, the intern still needs management and polishing. But the days of spending big bucks to get to page 1 of search results are coming to a close, that's going to eliminate a lot of jobs.
I have a friend who works as a real estate appraiser. They write these long reports that are often very boilerplate. This is a good application of AI - stuff that isn't meant to be personal or creative.
Like your plan. You can probably write the outline, and have the actual plan written by the AI.
Also, grammar checking is a really good application of AI, I would think.
Just please, let's not pretend this is a person writing it.
There was a really great quote from the instructor in my class "Treat AI like it's an intern who's work you need to review. You don't just shoot it out to the client or the senior management. You review the work, editing as needing, giving credit for the initial research as necessary to the intern. But you don't expect the intern to deliver a finished product."
Would AI change "who's" to "whose," I wonder?
Probably 😀 I can barely spell my own name without spell check these days.
Clown
Matt Stieb (probably) did not write that headline.
NYMag's content editor or whomever is in charge of clickbait titles wrote that headline.
Sidebar: Who still has a NYT subscription? I cancelled mine three weeks ago; Sulzberger is a walking advertisement as to why owners should have zero editorial control over content.
I gave mine up a few weeks ago as well. The way they handled all the Biden happenings and essentially give Trump a pass on all the stuff he states was the final straw for me. They should have a front page column every day called "What idiotic thing did Trump say today?"
My students (middle school social studies) are using AI and the results are absurd. The stuff they turn in is just transparently fake.
zoom's ceo is pushing the dumb as rocks idea that they're going to create AI zoom avatars that we all send to meetings as our substitutes and they'll all just talk to each other and then they will somehow do...something.
They're just trying to maintain relevancy. A lot of companies have banned the use of AI notetakers like OtterAI (which is annoying, but understandable). There is a cool use case for AI avatars though in training. Updating training videos is expensive and stuff changes really fast today.
I'm playing around a bit with this one https://www.heygen.com/ but the reviews I've seen on the translation feature were not promising -- at this time, I'd bet in a year they have that dialed in. Being able to just update a training script without having to rerecord would/will be amazing.
I have a bunch of meetings where I basically just listen while people rattle off their updates, I'm liking the idea of being able to just read a transcript or have an avatar read me the highlights while I'm doing other tasks that require minimal brain power.
Google has been trying to destroy its own search engine for over a decade and they've finally found a way to do it with AI. Today, I searched for a particular coffee which the brand turned up but the wrong roast -- over and over and over. Until one of those alternative search choices came up to search for a review of the roast I was searching for. Clicked that. First thing it offered up was an article on Star Trek, and then Tesla, and then quantum computing and then who the hell cares anymore? Google has totally screwed the pooch. Its search results, even if they are on point, cannot be trusted. And if you can't trust the results from a search engine, it really isn't a search engine anymore, is it?
Have you tried Kagi? I'm still getting used to it, but it really cuts through the 'sponsored links' crap.
I think this is a case where the hyperbolic headline is probably accurate, that ad is getting roasted all over the place, I don’t think I’ve seen a single person anywhere say it wasn’t horrible.
My concern in viewing this ad was how tragic it was that a) the little girl wouldn't learn how to write herself, and that b) the parent seemed very encouraging/helicopter-y by suggesting the use of AI and even providing a prompt. No need for the little girl to think for herself.
And the rise of AI overlords and all that, too.
The ad is the sort of thing only an SV marketing droid could think was a good idea. But this sort of thing is going to get much worse, I think.
The HugeCo that owns my employer also owns a captive Indian outsourcing company, and we effectively have to use them. Work quality varies from quite good to irrelevant trash, mostly on the dimensions you expect - accurately communicating complexity is the main, but not the only, problem.
Enterprise Autocomplete enters stage left.
Lots of folks there are using it to communicate with us now. One team just lost close to a month because the robot generated word-slop that convinced the team lead everyone was on the same page when they were emphatically not, and the remote team built something that kind of vaguely looked like one part of the spec.
That made me extremely aware that we effectively have to live with communicative intermediaries now, just like we live with financial intermediaries[1]. But because human communication is vastly wider than a dollar figure, there is much more opportunity to, um, "add value" in all sorts of ways of which we've barely scratched the surface.
I have no idea what the stable state of this looks like, it will take us some time to get there. One thing I do constantly now is try to determine to what extent I'm talking to a robot. In some cases when it doesn't seem rude and I can't figure out what's going on I've started asking for the prompt they used - you can see what they actually thought they should be talking about.
What a weird world.
[1] I don't know what the percentages are, but many US people don't personally exchange money for most things they "buy". They contract with a third party to handle payment and settle up later. And most of those who don't use credit cards use debit cards that outsource the money-holding and transfer part.)
*** Spoiler from 2013 *** 😀
This really reminds me of the service that Theodore provided in his job in "Her". Not just having someone write a letter for you but to have them become the main intermediary with your loved ones for YEARS. The dystopia is running just under the service while above the surface the adults are clearly unequipped to deal with life.
> without anyone having to bother making any effort
The little girl made an effort when she asked the AI to write the letter, and when she put the letter in an envelope with a stamp on it. (Or when she clicked on the email 'send' button.) Similarly, the athlete had to make an effort to tell the AI to create a response for .
Does he have to do a poll? Google is trying way too hard to make AI make sense when it doesn't make sense. And saying "everybody hates this" is rhetorical and you know it. It isn't intended to be a quantitative statement.
And then it was gone: https://www.cnbc.com/amp/2024/08/02/google-pulls-ai-ad-for-olympics-following-backlash.html