Fedi, can you help me find real world examples of Bad (meaningless, misleading) AI generated Alt text!
What the fuck. Stop finding ways, nature. At least not into the living room.
#IcelandAir website is basically unusable on Firefox for me. The "My Booking" menu item simply doesn't work, and all the blur effects make the site too slow to use at all.
Apparently IcelandAir thinks it's fine and dandy to be so incompetent as to only target a single browser.
If they are that incompetent regarding their website, I wonder how competent they are regarding anything else?
Oh, and they don't accept "name+icelandair@example.com" e-mail addresses. Their check-in system auto-corrupts them to "name plus icelandair@example.com". 🤦♀️
Yes, they have autocorrect enabled on an e-mail field, somehow. Yes, it's the website, nothing on my end is doing any of that.
Absolute nincompoops.
... the funny thing is, I know enough about how the search core is built to make an educated guess at how this happened.
Search core farms the query out to (possibly dozens of) completely independent software engines to answer the query. Those engines have their own backing datastores and they don't synchronize to each other; it's necessary they don't to make search fast. So it's entirely possible that the Top Stories core is pulling data that is more recent while the AI Overview core has built its information off of an older search result cache (and possibly hot-cached this answer before the NYT confirmed the death when the news broke locally / on social media). Google, as an institution, leans on the tradeoff "eventual consistency is good enough," so this kind of bug is seen as a predictable side-effect of the chosen tradeoffs.
Does that make us feel better?
No, nor should it. Get your shit together, Google.
ajuvo ✔
boosted
East Berlin, 1990 or 1991
East Berlin, 1990 or 1991
Mais, résoudre les problèmes concrets n'ai pas le rôle du gouverneur... Son rôle c'est de se faire prendre en photos, où il fait semblent de résoudre des problèmes, et de les publier sur les réseaux sociaux...
“Pas une polémique, mais un appel” : le message Facebook du doyen de la Faculté des Sciences de Sfax fait réagir !
deutrino
boosted
> Google is using AI age checks to lock down user accounts https://www.theverge.com/news/716154/google-ai-age-estimation-under-18
FTA: "The company says it will use the information a user has searched for or the types of YouTube videos they watch to determine their age. "
I may need to stop watching #SesameStreet and cartoons clips on YouTube
#tech #fuckery #Google #fail#CivilRights #privacy #censorship
> Google is using AI age checks to lock down user accounts https://www.theverge.com/news/716154/google-ai-age-estimation-under-18
FTA: "The company says it will use the information a user has searched for or the types of YouTube videos they watch to determine their age. "
I may need to stop watching #SesameStreet and cartoons clips on YouTube
#tech #fuckery #Google #fail#CivilRights #privacy #censorship
Now Docker wants me to sign in to use their AI. They don't even try to make it attractive with a list of things it can help with, it's just the first thing you see when you open the desktop app now. #fail
Le réseau national universitaire tunisien, qui héberge les sites web de tous les établissements universitaires, a subit des attaques ces dernières semaines... Pour la contrer la décision vient de tomber "Les sites web hébergés au CCK ne sont accessibles actuellement qu’à partir du territoire Tunisien."
Alex Akselrod
boosted
Judging by Trump’s past few months in office it’s easy to see why he went bankrupt 7 times and was seen as the biggest laughingstock in the business community.
Judging by Trump’s past few months in office it’s easy to see why he went bankrupt 7 times and was seen as the biggest laughingstock in the business community.
gittaca
boosted
I took a survey today for Masterclass.com, a product I generally like and support. They appeared to be leaning into AI, which I detest. I thought about it after, so emailed them that if they started offering classes on how to use AI, I was probably going to cancel my membership, because I find it morally reprehensible.
I got an auto-respond email from an AI Bot offering to help me cancel my subscription. There you go.
hamish campbell
boosted
⁂ Article
Can This Platform Survive? Governance Challenges
papers.ssrn.com
Can This Platform Survive? Governance Challenges for the Fediverse 8
A paper on the Fediverse by Thomas Struett, American University – School of Communication, Aram Sinnreich, American University – School of Communication, Patricia Aufderheide, American University – School of Communication, Rob Gehl, York University.
Interesting #mainstreaming look, that bypasses the grassroots it’s actually talking about, this is a common issue with academic writing, am at Oxford this […]
I took a survey today for Masterclass.com, a product I generally like and support. They appeared to be leaning into AI, which I detest. I thought about it after, so emailed them that if they started offering classes on how to use AI, I was probably going to cancel my membership, because I find it morally reprehensible.
I got an auto-respond email from an AI Bot offering to help me cancel my subscription. There you go.
WTF, Rogers internet wifi?
Where's the time machine to travel back to 1969 to access #wifi before the internet existed. #Corporate#Fail
Tim Chambers
boosted
"AI models incorrectly answered more than 60 percent of queries about news sources."
"Perplexity provided incorrect information in 37 percent of the queries tested, whereas ChatGPT Search incorrectly identified 67 percent (134 out of 200) of articles queried. Grok 3 demonstrated the highest error rate, at 94 percent."
