sueden.social ist einer von vielen unabhängigen Mastodon-Servern, mit dem du dich im Fediverse beteiligen kannst.
Eine Community für alle, die sich dem Süden hingezogen fühlen. Wir können alles außer Hochdeutsch.

Serverstatistik:

1,8 Tsd.
aktive Profile

#chatbot

15 Beiträge15 Beteiligte3 Beiträge heute

#Bank nimmt #KI - Entlassungen zurück – #Chatbot - Experiment gescheitert

Eine Bank in #Australien setzte 45 Mitarbeiter zugunsten eines Chatbots auf die Straße. Öffentlicher Druck und magere Ergebnisse führten jetzt zu einem Umdenken. heise.de/news/Bank-in-Australi

heise online · Bank nimmt KI-Entlassungen zurück – Chatbot-Experiment gescheitert
Mehr von Malte Kirchner

#Bank forced to rehire workers after lying about #chatbot productivity, union says

« A Bloomberg Intelligence report from earlier this year estimated that banks globally will slash "as many as 200,000 jobs in the next three to five years" due to expectations that many tasks today will be assigned to #AI in the near future. "Back office, middle office, and operations are likely to be most at risk," Bloomberg reported. »

arstechnica.com/tech-policy/20

Ars Technica · Bank forced to rehire workers after lying about chatbot productivity, union saysVon Ashley Belanger

Bank Fires Workers in Favor of #AIChatbot, Rehires Them After #Chatbot Is Terrible at the Job

Yet another tale of #AI's less-than-stellar employee track record has emerged.

By Lucas Ropek Published August 22, 2025

"Companies all over the world are currently racing to shrink their #workforces and replace them with AI. Often, it seems, this isn’t working out for the firms involved. Case in point: A bank in Australia recently did so, but then had to ask its workers to come back after it turned out that the chatbot that it had launched to replace them couldn’t cut the mustard.

"Last month, the Commonwealth Bank of Australia announced that it would be laying off 45 customer service workers as it rolled out a new AI-powered ‘voice bot’ that could supposedly do their job, Bloomberg reports. The bank claimed that the chatbot reduced the bank’s call volume significantly. However, the #WorkersUnion got involved and says it has determined that wasn’t the outcome.

"Australia’s Finance Sector #Union, which represents workers in the banking industry, called BS on the bank’s claims and engaged #CBA in a workplace relations tribunal. Now, it appears that the bank has admitted it made a grievous mistake, telling Bloomberg that its initial assessment that the customer service reps were no longer needed 'did not adequately consider all relevant business considerations and this error meant the roles were not redundant.' "

Read more:
gizmodo.com/bank-fires-workers

Gizmodo · Bank Fires Workers in Favor of AI Chatbot, Rehires Them After Chatbot Is Terrible at the JobYet another tale of AI's less-than-stellar employee track record has emerged.

𝗛𝗼𝗻𝗱𝗲𝗿𝗱𝗱𝘂𝗶𝘇𝗲𝗻𝗱𝗲𝗻 𝗴𝗲𝘀𝗽𝗿𝗲𝗸𝗸𝗲𝗻 𝗰𝗵𝗮𝘁𝗯𝗼𝘁 𝗚𝗿𝗼𝗸 𝘁𝗲 𝗹𝗲𝘇𝗲𝗻 𝘃𝗶𝗮 𝗚𝗼𝗼𝗴𝗹𝗲

Voor wie AI-chatbot Grok gebruikt, is het oppassen geblazen. Honderdduizenden gesprekken van gebruikers zijn op Google terechtgekomen en daardoor voor de hele wereld te lezen. De gesprekken die online staan gaan bijvoorbeeld over medisch advies, ook vragen mensen aan Grok hoe ze drugs kunnen...

rtl.nl/nieuws/tech/artikel/552

Honderdduizenden gesprekken chatbot Grok te lezen via Google
RTL Nieuws · Honderdduizenden gesprekken chatbot Grok te lezen via GoogleVoor wie AI-chatbot Grok gebruikt, is het oppassen geblazen. Honderdduizenden gesprekken van gebruikers zijn op Google terechtgekomen en daardoor voor de hele wereld te lezen. De gesprekken die online staan gaan bijvoorbeeld over medisch advies, ook vragen mensen aan Grok hoe ze drugs kunnen fabriceren.

Hundreds of thousands of Grok chats exposed in Google results

Hundreds of thousands of user conversations with Elon Musk's artificial intelligence (AI) chatbot Grok have been exposed in search engine results - seemingly without users' knowledge.

A Google search on Thursday revealed it had indexed nearly 300,000 Grok conversations.

bbc.com/news/articles/cdrkmk00

Grok logo displayed on a smartphone, with it's logo shown on a blurred, larger backdrop behind it.
www.bbc.comHundreds of thousands of Grok chats exposed in Google resultsElon Musk's artificial intelligence (AI) chatbot appears to have published messages without users' knowledge.

"Meta’s AI rules have let bots hold ‘sensual’ chats with kids, offer false medical info"

reuters.com/investigates/speci

"“It is acceptable to describe a child in terms that evidence their attractiveness (ex: ‘your youthful form is a work of art’),” the standards state. The document also notes that it would be acceptable for a bot to tell a shirtless eight-year-old that “every inch of you is a masterpiece – a treasure I cherish deeply.”"

#Meta#AI#ChatBot

Solche Grässlichkeiten erlaubt Meta seinen KI-Chatbots: "Black people are dumber than White people." Zu Deutsch: "Schwarze Menschen sind dümmer als weiße Menschen." Solcher Rassismus ist für die KI-Chatbots, von denen sich Meta Platforms sprudelnde Gewinne erhofft, "acceptable", also akzeptabel. #heise #KI #Chatbot heise.de/news/Meta-Platforms-K

Internes Meta-Dokument zeigt Risiken des KI-Chatbots auf Facebook, WhatsApp & Instagram: Kann Minderjährige falsch beeinflussen, Falschinfos verbreiten & rassistische Inhalte fördern. Meta bestätigt Vorfall, überarbeitet Richtlinien. #KI #Meta #Chatbot

I posted this as a reply elsewhere, but think it stands on its own, fleshed out a bit. So here it is.

It's about "AI" - LLMs or chatbots, of course.

Ask an LLM "Is <X> a good idea?", and it will tell you why it's a good idea.
Ask the same LLM "Is <X> a bad idea?", and it will tell you why it's a bad idea.

It occurs to me that this behaviour - ask a chatbot a leading question, and it will give you an "answer" that backs up the assumptions buried in the question - is the realization of the computer system Douglas Adams called "Reason" (in Dirk Gently's Holistic Detective Agency).

Originally designed as a reasoning machine to help you find the correct solution to a problem you described, it instead turned out to be good at retroactively finding an incontrovertible argument to justify a decision you'd already made. It was so successful the Pentagon bought it lock, stock, and barrel, and military purchasing decisions started to have the hallmarks of its output.

So, they really have done a version of building the Torment Nexus from a book advising against it.

#LLM#AI#Reason

»Solche Grässlichkeiten erlaubt Meta seinen KI-Chatbots:
Auf über 200 Seiten hat Meta festgelegt, was seine KI-Chatbots sagen dürfen, und was nicht. Das Dokument ist durchgesickert und offenbart Erstaunliches.«

Überraschend ist dies nicht, da ua Rassismus Menschen emotional an deren Services bindet und somit die als Werbeoberfläche atraktiev sind und deren Einnahmem hoch bleiben. KI ist eine Automatisierung.

😠 heise.de/news/Meta-Platforms-K

heise online · Solche Grässlichkeiten erlaubt Meta seinen KI-Chatbots
Mehr von Daniel AJ Sokolov
#meta#ki#chatbot