Canada

Last month, different country, same company:

https://apnews.com/article/australia-ai-errors-deloitte-ab54858680ffc4ae6555b31c8fb987f3

AI should never be used to compensate for laziness, incompetence or stupidity. If you aren't capable of doing your own job without a crutch, then you shouldn't be doing that job.

I got a cool concept... How about we stop subcontracting shit companies that use AI and let them go bankrupt?

Just don't hire "Toilet and Douche", they are corrupt.

AI vendors and AI-based companies are taking advantage of an often-overlooked asymmetry in how value and risk work.

MBA brain thinks: task makes us $500, has a 1% chance of making a mistake that costs us $200. So expected value is: 500-200x0.01, or $498. So if I can use AI to do task twice as fast, it’s all upside!

Reality is: value-adds tend to be situational, and risk tends to be systemic. If you do task x2, you might not see any actual extra profit as a result — due to dependencies, diminishing returns, and maybe supply and demand, that value might just fizzle out.

But if you make a small mistake, and then someone else makes a small mistake while building on top of your work, and so on… that risk compounds quickly and spirals out of control.

Automation like this should only be for approval. Anything it thinks needs to be denied then has to go to professional humans to see what nuance is there.