Generative AI is Like the Internet, Which Was Never a Solution Looking for a Problem
The Internet is/was a Tool that Enabled Many Solutions to Address Many Problems
There is an increasing number of voices out there that I feel misunderstand what’s going on with generative AI and large language models (LLM). It’s the hottest topic on the Internet now and for good reason. Rarely do we see something so amazingly different and capable emerge; but we need to get used to an increase in the pace. It’s awesome and scary at the same time, and there are people that are taking advantage of those emotions in order to promote their own brand. Some of this is driven purely by dollars, others by vanity (and ignorance).
Nothing ever seems to change, regardless of what changes around us. We’re human after all.
On the one hand we have people lamenting the fact that the word intelligence is used in the name. On the other hand we have people analyzing the inability of open source copies of LLMs to defeat the incumbents, which therefore must void Clay Christensen’s Innovator’s Dilemma story. That story already had some issues regarding disruption.
"All models are wrong, but some are useful." ~George Box (British Statistician)
When we try to fit things into our existing mental models, we’re essentially looking backwards.
Maybe we should call artificial intelligence (AI) artificial processing and analysis (APA). Is it intelligence? Not really, but no one was too worried about the use of this term until recently. Why is that? Quite possibly because they had imagined that it would manifest itself in some ‘2001: A Space Odyssey’ way? That hasn’t happened…yet.
More importantly, Generative AI is a tool, not a solution looking for a problem. As a tool, it has the potential to enable, accelerate, and integrate the development of many new solutions that help people get higher-context jobs done, more completely, better, and with fewer visible features. This is where the true competition exists, not with simpler low-end of the AI market disruptors somehow killing off the incumbents at the bottom of the market.
That’s not how it works. Christensen’s description was compelling, but stories about low end car manufacturers rising from the bottom end of the market wasn’t a compelling innovation story from a consumer perspective. Yes, they got cheaper cars…but they got the same job done, not a higher-context job. This was a production model improvement that was not visible to consumers in any way other than price and quality. It was painful for incumbents, but it was hardly disruptive. We’re still getting from point A to point B the same way - using roads - and we’re buying cars primarily through a dealer network. Where’s the disruption? Electronics have brought some integration of the struggle stack, but that’s about it.
Think of it this way, AI has been used to some degree in the new music platforms to predict what we want to hear. This was folded into a delivery platform that eliminated our need to buy multiple pieces of hardware and source our music in physical stores. It leverage other tools and technologies like the Internet and streaming to automate this process for music enthusiasts. The capabilities of the Internet and AI made it possible to devise a disruptive business model through consolidation, delivery and pricing.
But what didn’t happen was open source competitors for streaming disrupting any incumbent streamers. There was no new Internet created at the low end of the “Internet” market that rose up and defeated the big Internet. That’s imply not how to look at this. We’re seen continuous improvement but nothing truly disruptive at that level. Th solutions built with these tools are what disrupted markets - I’m thinking of you music industry, and you sellers of server-room technology.
My expectation is that we’ll see great things arise from these new tools. Possibly more tools built on top. Definitely new business models. I also expect see many things that should have never been made (like the Pet Rock), that will quickly fail because they don’t address and specific problem or use case and were based on some founder’s idea. What a waste of capital.
A group of clever people invented a new tool that helps people process significantly more information, faster, and from significantly more angles. That’s not a job-to-be-done, it’s also not a single solution. It’s an enabling technology that will help us get many jobs done faster, better, and more accurately.
🔥🔥 Now stop analyzing this stuff and go solve a problem with it! 🔥🔥