[Video] Ignore Buzzwords and Pay Attention to Attacks on Software Developers
Video download link | md5sum 38762e78b86e2ad566986088316cdafa
Chatbot Hype Waning
Creative Commons Attribution-No Derivative Works 4.0
THIS is the eighth and last video in yesterday's batch. It's about the "Hey Hi" (AI) hype that's already faltering, even based on mainstream media, which says shareholders are running out of patience as they don't see returns, only excuses and procrastination.
AI in the Machine Learning sense is nothing new. It's very, very old. Some of us did that decades ago. "The computing was not up to the software at that time though," one person recalled. "Now it has turned around and the software is a failure."
Yes, on all levels: text, code, images, video. Basically... it's rubbish and it is also a legal liability because they try to make plagiarism seem like "fair use" by automation and bulk processing, i.e. doing the same thing over and over again en masse. It's bad for the planet, it harms privacy, and it does all sorts of other bad things.
There are many lawsuits and almost no success stories, i.e. companies that are actually profitable in this area (investment being attracted is not the same as profit). Cory Doctorow wrote about it last December.
The "Hey Hi" (AI) hype won't die overnight (we've called it "HEY HI" for about 3 years already) because the discipline is so old and people are meant to think some kind of revolution happened in later 2023 because of chatbots (LLMs). They don't even qualify as "Hey Hi", "stochastic parrots" is a more suitable name for them.
Judging by media coverage, we shall see waning press coverage about this subject, i.e. the same as we saw with Web3, Blockchain, Metaverse, and Vision Pro. We saw amusing IBM comments about it last week, just like comments from Microsoft insiders (sometimes the press quotes them anonymously). They're deeply concerned the companies tie their fate to this buzzword or buzzphrase. It won't end too well as the general consensus is that the hype is over, so "let's move on..." (and make real products).
"The damage from conflating LLMs with AI is real," an associate tells us. "The damage from real world (mis-)use of LLMs is just beginning and hasn't peaked yet. Hopefully civilization / society can survive through the peak and beyond."
As I explain in the video above, even Google is sort of admitting that LLMs don't work. Because, to the company's credit, it did cull a lot of chatbot-generated or 'AI' [sic] sites. Frankly, we need to make those sites fail and vanish because they ruin the Web. It's ruining what's left of the Web and it's like weed growing in your garden.
"That problem is just beginning and Google can't keep up," an associate notes. "That's part of the conundrum."
In the next few posts we'll deal with subjects all this hype serves to distract from, ranging from financial perils to the bubble of software patents waning in the US court after Alice/35 U.S.C. § 101 precedents/changes (at SCOTUS 10 years ago). The U.S. Patent and Trademark Office (USPTO) is still granting many patents, but 'monetising' them isn't easy anymore. "Goodwill" won't pay the bills (or salaries) either. We'll discuss this in the next article, but some is covered in the video above. █