| ▲ | varispeed 3 days ago | |
The funniest part isn’t that AI hasn’t delivered profits. It’s that the only “value” most people got from LLMs was accidentally rediscovering what Google used to be before it turned into an ad-riddled casino. Executives mistook that novelty for a business revolution. After years of degraded search, SEO spam, and “zero-click” answers, suddenly ChatGPT spat out a coherent paragraph and everyone thought: my god, the future is here. No - you just got a glimpse of 2009 Google with autocomplete. So billions were lit on fire chasing “the sliced bread moment” of finally finding information again - except this time it’s wrapped in stochastic parroting, hallucinations, and a SaaS subscription. The real irony is that most of these AI pilots aren’t “failing to deliver ROI” - they’re faithfully mirroring the mediocrity of the organisations deploying them. Brittle workflows meet brittle models, and everyone acts surprised. The pitch was always upside-down. These things don’t think, don’t learn, don’t adapt. They remix. At best they’re productivity duct tape for bored middle managers. At worst they’re a trillion-dollar hallucination engine being sold as “strategy.” The MIT study basically confirms what was obvious: if you expect parrots to run your company, you get birdshite for returns. | ||
| ▲ | xpe 3 days ago | parent | next [-] | |
Grain of truth acknowledged, though much exaggerated -- to the point of forgetting history. > No - you just got a glimpse of 2009 Google with autocomplete. Go look at the kinds of "answers" you got in 2009 with Google's autocomplete. Time to drop this convenient nicely-packaged narrative. > These things don’t think, don’t learn, don’t adapt. Perhaps some don't -- for some intentionally constructed definitions of these terms. Adding a feedback loop (chain of thought, etc) challenges even such definitions though. > if you expect parrots to run your company, you get birdshite for returns. If this is a reference to "stochastic parrots" metaphor, then I would encourage you to read these articles and HN comments below, which push-back against misunderstandings or overuse of that metaphor: [1]: https://www.lesswrong.com/posts/HxRjHq3QG8vcYy4yy/the-stocha... [2]: https://www.lesswrong.com/posts/7aHCZbofofA5JeKgb/memetic-ju... [3]: https://news.ycombinator.com/item?id=44299996 [4]: https://news.ycombinator.com/item?id=44926540 [5]: https://news.ycombinator.com/item?id=44967655 [6]: https://news.ycombinator.com/item?id=43676755 | ||
| ▲ | kmarc 3 days ago | parent | prev [-] | |
While you are being downvoted here, I would like to express how much I liked how this thought summarized some of the pain points I have to deal with: (regardless of AI): > they’re faithfully mirroring the mediocrity of the organisation This is happening with to technologies / tools / etc... | ||