NVIDIA: As growth slows, Market will correct evaluation of AINVIDIA continues to announce reports of a growing business, but this growth's velocity is slowing down. The chip designer has reached peak interest on the financial market both in institutional and private investors, for which the buy-back plans of NVIDIA might be the only remaining driver of evaluation.
The delay of Blackwell will lead to a stall of growth for the whole business, as data centers establishing new circuits for AI business rather pre-order the newest generation of chips and wait out their expansion than make do with the current set - after all, this hardware is supposed to run for years and rushing it won't help the bottom line when customers asked for most efficient and most recent cutting-edge hardware.
Known problems with generative AI aside, resistance against training AI with the internet is moving forward. X, formerly Twitter, excluded web access to it's data treasure ever since Elon Musk took over, and other websites are to follow. Internet service providers and web services announce a significant rise of traffic and requests - according to their own identification: AI training bots scraping the internet for training data. The problem: these bots aggressively scrape even for pages the web service providers don't want any bots to see. Google established the standard of putting a simple text file called robots.txt with a list of pages you don't want in the index up onto the top of your website. The new AI training and scraping bots ignore this text file and violate the privacy and consent of service and content providers. So what began with Elon Musks attempt to make a profitable business out of the internet's once most favored website, is now becoming a movement in size similar to the movement of blocking ads. Large content providers will take measure against illegitimate scraping or AI bots themselves, so the largest database of training data will soon vanish away from sight if AI bot engineers won't pick up the fight against measures counterfeiting their product.
To see how the fight might end, one can refer to the war against Ad Blockers, which is in a similar bind of endless fights. Google, the most dependent business on advertising, would rather have them out sooner than later, and even manipulates the most favorized browser in the world to render ad blockers ineffective. However, any measures taken by Google to counterfeit the users' legitimate measures to select the content they want to see as well as protecting their computer from malicious intent lead to Ad Blockers coming out even stronger in their function, ending Google partially giving up the fight.
For AI, training with the applied and collective knowledge of the internet could become much more expensive as important content providers such as Stack Overflow, a large Q&A solutions database for engineers, might want to cash in on the AI industry early-on. In general, the quality of future AI models is not only endangered by a degression of quality (by ingesting AI results and being unable to tell the difference) but also by a degression of free access.
The AI expansion in the industry is underway, but as the generative nature of it, with all its disadvantages, as well as upcoming difficulties, whether in court or in the web, putting brakes on the growth of the industry, AI products might take longer to establish. AI systems in its current generation are limited to assisting in service; they're unable to creative or critical thinking and make the best impression on people who are easily impressed and lack of deeper understanding of what the AI generates. For a finished product of high quality, manual improvements and, in general, humans assisting the process are required. Purely AI-generated products directly distributed, due to common availability of the technology, already give the impression of being cheap, untrustworthy and probably something the internet generated, according to surveys of analysts in the public communications and marketing sector. The generative nature of AI products works well in products relying on repetition and a low intellectual approach, such as production of pop songs or any other action movie, but will come to show negatively on everything requiring deeper thought, such as the creation of tailored software solutions.
For the present, as long as AI is only the generative machine deriving a remix of everything humans once wrote down, AI will remain more to be an Assisting Intelligence rather than being Artificial Intelligence.
