@trochee@dair-community.social
Post #2908086
2025-06-18 05:12 UTC
Replies (3)
-
@aj@gts.sadauskas.id.au 2025-06-19 07:25
@trochee@dair-community.social There's also a lot of conflation that goes on under the umbrella "artificial intelligence". Over the past couple of years, a lot of software vendors that sell older machine learning tools noticed the hype around large language model sentence generators. So they've pulped all their marketing collateral around "big data", "predictive analytics", and "the cloud". Suddenly, all the hype for their existing product said "We're AI too"! A lot of enterprise automation and integration tools basically did the same thing too. "We're AI too!" Keep in mind these are platforms that people were talking about in conference keynotes I attended around 10 years ago. Other vendors have shoe-horned LLM sentence generator output as a source for their text entry or image upload fields. Think the email apps that use "AI generated" sentence generator output as a subject line. Or the content management systems that can create a page based on sentence generator output text. "We're AI too!" Heck, if the Windows 3.1 wizards were being sold today, they'd probably be marketed as "AI agents", even though it's all vanilla procedural code under the hood. So you end up with four or five very different types of tools that work very differently under the hood all being marketed as "AI". And then senior managers with no sense of curiosity are treating them all as if they are the same thing. As if these are the magical AIs from sci-fi films. The now ubiquitous "AI strategy". To the delight of software vendor salespeople everywhere, senior executives are considering automation/integration software, machine learning/big data tools, as if these tools are all the same thing: The AI characters from sci-fi movies. Instead of deferring to the IT or data management folks about whether or how, say, ML data analytics tools should be deployed, the edicts are coming down from the boardroom. Much to the relief of vendors, the pesky questions data scientists might ask of ML/data analytics tools, around data integrity and quality, are no longer being asked. Because the CEO has decided to hire C3PO. And with large language models, the senior bosses often don't have the experience to distinguish the outcome of work from the plausible-looking output of a sentence generation app. Add in the conflation with other tools, and there's a vast overestimation about what sentence generation bots like ChatGPT can actually do. Because to a lay person, the term "AI" refers solely to the sentence generators like ChatGPT. So when you call your data integration software an AI, in the lay executive's mind, it's the same thing as the sentence generators.
-
@violetmadder@kolektiva.social 2025-06-20 14:41
@trochee@dair-community.social Bet you ten bucks it couldn't figure out the way out of a wet paper bag.
-
@sascha_wolfer@fediscience.org 2025-06-21 05:53
@trochee@dair-community.social New turing test just dropped.