AI and the hype cycle

By November 21, 2026Business, Technology

American Journalist Cory Doctorow believes that much of the hype around artificial intelligence (AI), is solely being driven by the same people who are flogging AI, and this is solely to drive up the share price of their companies, which are a very long way from seeing a return on their massive investments in AI1.

The reason I bumped into the Doctorow comment above was because a friend of mine put me onto something called the Gartner Hype Cycle2. It is all about discerning the difference between hype around new technologies and what is commercially viable. The cycle comprises the following:

  • The Innovation Trigger: A breakthrough that provides a technological innovation. Often there are no usable products and commercial viability is unproven.
  • Peak of Inflated Expectations: Early publicity produces a number of success stories, often accompanied by scores of failures, which you tend not to hear about.
  • Trough of Disillusionment: Interest wanes and implementations fail to deliver; there is a culling of the companies promoting the technology. Investment continues only if the surviving companies improve their products.
  • Slope of Enlightenment: More instances of how the technology can actually benefit enterprises become more widely understood. Second and third generation products appear.
  • Plateau of Productivity: Mainstream adoption starts to take off. Criteria for assessing provider viability are more clearly defined. The technology’s applicability and relevance become established2.

My friend said that “Some people criticise the model as the timelines between phases are based more on analyst sentiment rather than empirical models or statistics which I would agree with. That said, in my experience the phases new technologies go through are absolutely spot on.”

Like many people, I was surprised and astonished at the November 2022 release of ChatGPT a ‘generative artificial intelligence chatbot’ developed by OpenAI. It uses ‘large language models’ specifically ‘generative pre-trained  transformers’ (GPT) which can generate text in response to user questions. That was the innovation3.

I first bumped into this in early 2023 when editing a paper on the history of opal mining in Queensland, and when trying to find out the source of the ‘opals are bad luck’ myth. With inflated expectations, I asked ChatGPT whence it came and it gave me the same old drivel that the source was Sir Walter Scott’s novel ‘Anne of Geierstein’ of 1829. This was despite the ‘bad luck’ drivel seemingly beginning some 60 years later4. Hardly a commercially viable technology when it spits out the same old drivel.

A second instance a week later was where I asked ChatGPT to give me a rundown of the stratigraphy (the succession of sedimentary rock units) of a particular sedimentary basin in the Northern Territory. It gave me a completely incorrect answer5. Again hardly a commercially viable technology. This was when I hit the trough of disillusionment.

Just short of two years later (March 2025), I got into a conversation with a person who writes computer code (in Python) as part of their job. Given the less than stellar results from my previous queries on the stratigraphy of the Northern Territory sedimentary basin, they tried assorted AI systems (ChatGPT, DeepSeek, Grok) and they all turned out abject rubbish6. I was still in the trough of disillusionment.

Geology was seemingly far beyond these AI systems, but I persevered, asking ChatGPT non-geological and mostly non-scientific questions. This is where I am now, with ChatGPT providing kick-starts to some of my rants. It effectively does the primary research for each rant. However, I make certain that I validate every assertion from one or more sources before I start writing. I still don’t trust it not to bullshit me, so I check everything. The irony in this article is that I did not use ChatGPT to research what I was going to write. Whether I have reached the plateau of productivity remains to be seen.

Sources

  1. https://www.abc.net.au/listen/programs/conversations/cory-doctorow-artificial-intelligence/107388236
  2. https://www.gartner.com/en/research/methodologies/gartner-hype-cycle
  3. https://en.wikipedia.org/wiki/ChatGPT
  4. https://www.blotreport.com/2023/04/15/the-opal-myth/
  5. https://www.blotreport.com/2023/04/23/dead-wrong/
  6. https://www.blotreport.com/2025/03/24/querying-machines/

One Comment

  • clive pegler says:

    admittedly i originally saw this as a meme (& note, i’m paraphrasing). “ask AI to discuss a topic one is unfamiliar with & it will amaze one, but ask it to expand on a topic in your field of expertise & it will expose itself as a fucking idiot ” nb 2: that was an industrial scale paraphrasing but in my infrequent & brief interactions, it seems to hold as reasonably accurate. Mind you, all it is doing is vacuuming up anything & everything ever written down factual or not, peer reviewed or not, so it is bound to spew out bollocks. Aditionally, i’ve been told, they have been trained, from the tenor of the questions, to attempt to glean what the questioner WANTS to hear.

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.