An AI Innovation Case Study You Didn’t Expect

Meta’s Chief Scientist has provided us with a case study that both perpetuates and exposes many myths about innovation.

It was the headline that grabbed me: “ChatGPT is ‘not particularly innovative,’ and ‘nothing revolutionary’, says Meta’s chief AI scientist”

It came through Refind (a great service if you’ve never heard of it), which recommends articles based on my interests. Clearly, this was tied to the word innovation. It was correct to make this recommendation.

The article features an interview with Yann Le Cun, the Chief Scientist of Meta (which is what Facebook once was). When asked about ChatGPT, the AI writing and search tool that has captured public attention, he said:

“It’s nothing revolutionary, although that’s the way it’s perceived in the public,” said LeCun. “It’s just that, you know, it’s well put together, it’s nicely done.”

This quote, the title of the article, and the interview got me thinking about the many myths and misconceptions of the word ‘innovation’ and has prompted me to clarify some of them that come through in this story.

Is ChatGPT Innovative?

The answer: It depends. My answer points to the first issue with innovation: context sensitivity.

For LeCun, he doesn’t see what OpenAI has done as particularly novel, given what he knows is in development at Meta, Google, and other places. He might be right. I need to learn more about the technical aspects of ChatGPT and the development market to assess its innovativeness. In his professional context, ChatGTP is one of many tools available that do similar things. He his making a comparison between models.

For the public, the answer is an emphatic yes. ChatGPT was the first mass-market, accessible model of AI-driven writing that captured widespread attention. Authors like Anne-Laure Le Cunff have compiled detailed taxonomies of hundreds of AI-powered tools used in creative production. Kirk Clyne, has written about how these tools support visual creativity and provided illustrated examples of their use. It’s not that ChatGPT was the first, the best, or the most accurate; it’s that it was the first that broke through into public consciousness. ChatGPT is the tool people use and experiment with on a massive scale. It likely won’t be forever, but it was the first. And as my doctoral research advisor once said about innovation: “people remember the first; they don’t always remember the best.”

This brings up another point: scale, scope, and influence. For example, ChatGPT has already massively disrupted education institutions as students and faculty struggle with how AI writing and production tools fit into learning. Artists are debating whether tools like DALL-E 2, Stable Diffusion, and other text-to-image tools are augmenting, promoting or stealing their work. These discussions are happening worldwide and on a massive scale.

We will soon see these conversations in the corporate world, government, and scientific communities before long.

Being Of Their Time

Tools like ChatGPT, DALL-E 2 and others might become the Netscape Navigator of AI (the one that was first but didn’t win out in the browser wars). Where we once declared Google the winner of that war, we might now reconsider. What happens if tools like ChatGPT replace Google with search? What if they end up killing the browser altogether? This is another myth and lesson in innovation: viewing something new in light of what is available now.

Browsers, search engines, and other tools we take for granted might be ‘of their time.’ They may only exist for a few years, supplanted by chat tools or something else that AI delivers. It doesn’t feel that way because of how networked they are in the context of our everyday lives. But just as VCRs, radios, and Walkmans other media tools captured our attention, they were all of their time.

Sometimes the best innovations aren’t even new; they are just new in context. Consider the role of home visits and in-home care and how it’s been an innovation in healthcare. This is a very old idea brought forth in a new context.

The stories we tell, the language we use, and the innovations we create are largely time-bound.

Where We Sit

Another innovation issue is always about where we sit in relation to it. As I mentioned earlier, Yann LeCun operates in a world where he always sees AI innovation.

“OpenAI is not particularly an advance compared to the other labs, at all,” said LeCun. “It’s not only just Google and Meta, but there are half a dozen startups that basically have very similar technology to it,” added LeCun. “I don’t want to say it’s not rocket science, but it’s really shared, there’s no secret behind it, if you will.”

LeCun has a vested interest in both diminishing the work of OpenAI (the maker of ChatGPT and DALL-E 2) and elevating the work of others. This is why we need to consider his innovation assessment from other perspectives. Most innovators are actively advancing an idea of change. This might lead to a perspective (bias) that ignores other possibilities that may have benefits unrecognized by their innovation. I can’t evaluate LeCun’s assessment of what Google and Meta do, but I know he’s invested in seeing his work advance. Right now, OpenAI is getting all the attention, which also matters.

Attention and Adoption, Risk and Reputation

LeCun said that one of the reasons OpenAI could deliver ChatGPT is because of its size. Companies like Google and Meta had too much risk attached to having their AI efforts fail. OpenAI, largely unknown to most people before last autumn, could try something and fail as the repetitional risk for them was low. This points to the myth that large companies- especially ones known as innovators- are good at or likely to innovate. This isn’t true. Resources need to balance with risk. This is where the ‘young and scrappy’ model of start-ups can deliver in ways that industry leaders need to.

Think about this with your organization: how can you design out some of the risk attached to trying something new?

Adoption and attention are essential factors as well. The interview with LeCun is partly meant to deflect attention from one innovation and illuminate others. While Meta and Google are well-known, they aren’t considered leaders in AI by the public. They are both known as the parents of Facebook or the masters of search. They are struggling to get people to see them differently — it’s one of the reasons why Facebook renamed itself. Open AI has the attention, which fuels the adoption of ChatGPT, generating more attention and adoption. It’s a virtuous cycle that creates scale and opportunities for scaffolding its popularity into something else.

Scaling and Scaffolding

The last point to raise about this innovation is to point to how OpenAI has used tools developed by others to create its product. This is one of the arguments for Open Source technology or shared-Intellectual Property. By sharing what they know, the companies involved in AI development can scaffold ideas to innovate. Most every substantive innovation was built on something else. It’s just as the often phrased quote about standing on the shoulders of giants suggests: we build on what’s come before us.

The myth of the lone genius (usually male, white, and middle class, too) is just that: a myth.

AI has been developing for forty years (or more), and the latest version of the tools we see today will look primitive in the years to come. Part of the innovation of these tools is not what the technology is, but what we humans do with it. How will we design our work? What will our creative value be in light of these tools? How can we learn more, better, and more deeply about the world with them? Or will we?

This is where the important innovations lie ahead. We’ll do that if we learn what the myths are and live not by them, but by what we need, want and experience. Those are the case studies to come, not just the ones about technology.

Image Credits: Cameron Norman, DeepMind on Unsplash,  Zac Wolff on Unsplash

Comments are closed.

Scroll to Top
%d bloggers like this: