Eine Umfrage zeigt, dass nur 2 % der kanadischen Unternehmen eine Rendite aus generativen KI-Investitionen erzielten

Oh wow, I love this article. It has everything I hate about AI articles. Here are some highlights:

> Only 2 per cent of Canadian companies are seeing a return on their investments in generative artificial intelligence despite widespread adoption by businesses, according to a new survey from KPMG. KPMG polled 753 executives in Canada and found that 93 per cent said their organizations are using AI in some way, up from 61 per cent a year ago. But only a handful said they are seeing benefits from a technology that is supposed to deliver productivity gains and cost savings.

Ok, that is some interesting data from a pretty reputable source. Let’s look at their analysis.

> “That 2 per cent was disappointing and surprising. I would have hoped and expected it to be higher,” said Stephanie Terrill, Canadian managing partner of digital and transformation at KPMG. **While generative AI can be prone to errors (known as hallucinations)**, the **reasons** for the low return on investment so far have to do with **how Canadian companies are approaching the technology**.

What? So, you are telling me it is prone to error – that is fine, all new technology has errors. But apparently that isn’t the reason executives aren’t benefiting from it. The real reason there hasn’t been benefits is we are using it wrong. It isn’t that the product is underwhelming, it is that we aren’t using the tool properly. This feels very much like „don’t piss on my leg and tell me it is raining“ type situation. It isn’t hallucinations, it is errors.

> Even with these challenges, three in 10 respondents to the survey said they expect their AI investments to provide a return within the year. The finding about lack of returns so far could add fuel to the criticism that generative AI tools – which can be used to write and analyze documents and presentations, handle customer service and employee inquiries, automate rote office tasks and more – are overhyped.

and

> Evidence of the benefits of generative AI has been somewhat mixed. A report from the Massachusetts Institute of Technology earlier this year garnered lots of media coverage because it found that 95 per cent of organizations are seeing zero return from their AI efforts. That finding has been held up as proof by AI skeptics that the technology is not all it’s cracked up to be.

I think AI could be truly powerful – much like the internet. But that doesn’t mean there isn’t a bubble. That also doesn’t mean it is inherently useful or productive or beneficial. I don’t think we aren’t using it properly, I think the technology is just extremely immature. That isn’t an *us* problem, this is a *technology* problem.

and

> Ms. Terrill also cited Canadian cautiousness as a reason for the low ROI number in the survey. “It’s a much more risk-averse culture, and thinking about the hallucinations when AI doesn’t work,” she said. “What about when it does work?” Given the productivity challenges facing Canada, Ms. Terrill said that companies should integrate AI into their operations even faster.

Yep. In the eyes of KPMG, the technology isn’t the problem. We are the problem. I hate this analysis.