"Immigration professionals and people applying to enter Canada say they're increasingly getting refusal letters they think don't make sense — leading them to wonder whether their cases are being fully and properly reviewed by a human being." #ai #aierrors #immigration #migrant #canpoli
Immigration lawyers concerned IRCC's use of processing technology leading to unfair visa refusals | CBC News
Immigration professionals and people applying to enter Canada say they're increasingly getting refusal letters they think don't make sense — leading them to wonder whether their cases are being fully ...
www.cbc.ca
September 16, 2025 at 4:34 PM
Everybody can reply
The McDonald's AI that served chaos? It's real! 🍔 Our blog post unpacks the comical drive-thru errors, exploring conversational AI and its unexpected hiccups. A must-read for tech enthusiasts! #AIInnovationsUnleashed #QSRAutomation #AIErrors
bit.ly/4mWMhrS
bit.ly/4mWMhrS
September 3, 2025 at 7:54 PM
Everybody can reply
A critical takeaway: LLMs introduce new classes of bugs due to their inherent limitations. Relying on them in critical systems via protocols like MCP demands rigorous error handling & validation beyond simple schema checks. #AIErrors 6/6
August 10, 2025 at 7:00 AM
Everybody can reply
🤖 AI in Disaster Response: Critical Errors Threaten Public Safety. Recent Russian earthquake highlights the risks
AI Deception: Russia Earthquake Errors
#ArtificialIntelligence #AIerrors #RussiaQuake
AI Deception: Russia Earthquake Errors
#ArtificialIntelligence #AIerrors #RussiaQuake
AI Deception: Russia Earthquake Errors
AI in Disaster Response: Critical Errors Threaten Public Safety. Recent Russian earthquake highlights the risks
techalarab.com
August 5, 2025 at 3:45 PM
Everybody can reply
1 reposts
I was told to ask Google Gemini if all numbers have an e in them. It says that _one_ , _ten_ , and _twenty_ don't have an e. When I corrected it Gemini doubled down then contradicted itself
#googlegemini #llm #ai #funny #aierrors
#googlegemini #llm #ai #funny #aierrors
August 3, 2025 at 12:26 PM
Everybody can reply
1 reposts
"When a judge’s ruling sounds like a chatbot's first draft, what’s the wildest legal jargon you’ve encountered? 🤖⚖️ #AIErrors #LegalTech #CourtroomChaos LINK"
July 25, 2025 at 6:18 PM
Everybody can reply
Errors found in US judge’s withdrawn decision stink of AI A US district court judge has withdrawn his decision in a biopharma securities case after lawyers noted that his opinion referenced fake quotes and other erroneous case... @cosmicmeta.ai #AIerrors
https://u2m.io/ABx7wEMc
https://u2m.io/ABx7wEMc
Errors found in US judge’s withdrawn decision stink of AI
It’s not just lawyers using it.
www.theverge.com
July 25, 2025 at 10:38 AM
Everybody can reply
Hacker News debated AI Overviews' unreliability, highlighted by Google falsely declaring Dave Barry dead. This sparked concerns over AI accuracy & Google's accountability, questioning the future of factual information online. #AIErrors 1/6
July 21, 2025 at 1:00 AM
Everybody can reply
💡 AI Mistook Chocolate for Medicine: Neural networks trained on drug databases confused cocoa for cough syrup. 🍫 Now doctors are prescribing hot cocoa for depression! ☕ #TechTrends #AIErrors Fansy cloths for women: cutt.ly/GrRd2il8
June 27, 2025 at 11:49 PM
Everybody can reply
1 likes
My latest on the way that A.I. models make mistakes about matters of fact that they have the knowledge to avoid. #AI #AIerrors #AJI #ArtificialJaggedIntelligence #AIhallucinations
AI should know this is wrong: the sharp edge of jagged AI
It's 2025 and hundreds of millions of people are using AI in their jobs and personal lives (source: The Conversation). Sadly, many of these people probably don't how badly wrong AI can go.
www.linkedin.com
June 18, 2025 at 6:30 PM
Everybody can reply
2 reposts
3 likes
#ai #aierrors
I gave #gemini pro 5 text logs of my workouts from my Hevy app. Simple Dumbbell kg x reps etc.
❌It added all 5 logs for total volume wrong.
✔️I manually calculated to check Hevy.
❌Spent 2hrs going thru logs with Gemini.
We've managed to make computers bad at basic maths tasks.
Not cool
I gave #gemini pro 5 text logs of my workouts from my Hevy app. Simple Dumbbell kg x reps etc.
❌It added all 5 logs for total volume wrong.
✔️I manually calculated to check Hevy.
❌Spent 2hrs going thru logs with Gemini.
We've managed to make computers bad at basic maths tasks.
Not cool
May 3, 2025 at 12:12 AM
Everybody can reply
🚨🔍 Ever heard of "vegetative electron microscopy"? It’s often misunderstood! 🤯 A 1959 paper mix-up has baffled AI and researchers, leaving a digital fossil. What other scientific myths have you encountered? #AIErrors #ScienceGoneWrong #DigitalFossils LINK
April 18, 2025 at 2:18 AM
Everybody can reply
Правильный ответ — 5228 (нейропомощник «Брейва» попытался меня надуть). Про функцию column/столбец не знал или забыл. #excel #exceltables #brave #aierrors #incorrectanswers #answers #printscreens #на_ду_ли
April 17, 2025 at 5:43 PM
Everybody can reply
LLMs aren't perfect! ⚠️ They don't truly "understand" language.
They rely on patterns from their training data. This can lead to "hallucinations" (incorrect answers). 👻
How do you handle AI hallucinations?
#AIlimitations #Hallucinations #AIErrors
They rely on patterns from their training data. This can lead to "hallucinations" (incorrect answers). 👻
How do you handle AI hallucinations?
#AIlimitations #Hallucinations #AIErrors
April 2, 2025 at 8:46 AM
Everybody can reply
ChatGPT isn’t taking over the world anytime soon 🙄🤷
#billionsofdollars #chatgpt #chatgpterrors #chatgptbugs #chatgptglitches #ai #aierrors #aiglitches #aibugs #errors #glitches #bugs #iancampbellstuff
#billionsofdollars #chatgpt #chatgpterrors #chatgptbugs #chatgptglitches #ai #aierrors #aiglitches #aibugs #errors #glitches #bugs #iancampbellstuff
March 2, 2025 at 5:17 AM
Everybody can reply
1 likes
Turns out FRE 702 is a “song” on the “album” “The Federal Rules of Evidence.” Lawyers 1, AI 0.
#AIerrors #stupidAI
#AIstupidity #ArtificialStupidity
#AIerrors #stupidAI
#AIstupidity #ArtificialStupidity
February 12, 2025 at 12:01 AM
Everybody can reply
1 likes
Hey, here's Copilot being all inaccurate on searches I did about how long journals should retain copies of a peer review. #aierrors
February 4, 2025 at 7:08 PM
Everybody can reply
»#Anthropic's new #Citations feature aims to reduce #AIerrors: allows its #AImodels to provide references to “the exact sentences and passages” from docs they use to generate responses.« https://techcrunch.c... #tech…
January 24, 2025 at 7:39 PM
Everybody can reply