For a long time, AI tools and AI development have been stuck in three areas:

1. Working quietly in the background (e.g., for creatives in Photoshop or for all of us in face recognition on images)
2. Performing expert tasks as a standalone product (e.g. Celonis, opseyes) or in academia (alphaFold)
3. Being a subject of research, especially around model architecture (see e.g. the rise of transformer models)
   

While my claim is in no way scientific, I would like to argue that with the publication of ChatGPT AI has in the winter of 2022-2023 leaped into a place which now actually can be considered disruptive innovation.

Disruptive innovation

Disruptive innovation is a term coined by Clay Christensen in his works beginning in the late 1990s. It is often times misunderstood and used to describe all kinds of innovation, which hinders many innovators to actually leverage Clay’s insights (as he argued himself: HBR). In brief, it can be understood as an innovation allowing an organization to attack a low-margin part of an existing market or a new market, gradually moving into more profitable regions and in the process of doing so slowly suffocating incumbent actors who–completely rationally–escape the competition in low-margin regions. If the disruption succeeds, either its methods will be adopted by the entire competition or it will push them all away. On a societal sidenote, Clay argued that disruptive innovation is one of the main engines in economic growth. On a more philosophical sidenote, I would argue it can also be an engine for degrowth or regenerative growth.

AI has become disruptive

With the likes of ChatGPT , AI-powered tools become widely availabe to support day to day tasks.
Already about a year before ChatGPT, Midjourney and Dall-E were the reason for an outcry in the arts scene. With these tools, everyone can write a description in natural language and can watch the tools create an image capturing that description. There is a question about copyright and the usage of artists’ work in the training data, I will touch on those later. For now suffice it to mention that these tools directly attacked the labour of artists. Why hire an expensive artist if you can spend a couple of dollars to let an AI generate the image you want? You even have much faster turnaround times and can iteratively close in on the perfect design.
Similarly for ChatGPT, you can replace many hours of copywriting and editing directly with an AI.

I argue that this is the first time AI really becomes a disruptive innovation in Christensen’s definition.
Art and literature are cultural pillars of our societies. They are also massive industries, with a combined annual global revenue of around $140 billion (source: Statista and wordsrated.com). Writing and painting are also essential parts of the production process in many enterntainment industries, most obviously in computer gaming and movies.
Imagine now a startup attacking the incumbents in one of these industries by massively leveraging AI-produced assets, arguably sacrificing artistic integrity for speed and low prices. That startup could invest a relatively higher share of its budget into marketing and the other parts of production, or it could aim for lower margins — either way, it would be attacking the incumbents in exactly the way outlined by Christensen.

This is the first time this really happened because so far AI was not yet ready to be used at scale. It powered smaller features, e.g. cropping of subjects in Photoshop or isolation of voices and instruments in digital audio software.
Uses of AI in de-aging, as employed famously by Disney in several Star Wars movies, and similar applications were very cost-intensive and can be seen as an instrument to differentiate: it is only sensible to use de-aging when you have Mark Hamill on cast and a Star Wars license to use.

Taking away jobs?

As a detour, I would like to address the question if AI will take away jobs.
Conventional wisdom has it that big technological breakthroughs create employment in the long run, not remove it. Short- and medium term however, and even more tangibly on a personal level, the word “disruption” is fitting. If entire lines of work become redundant, and if the people employed in these fields cannot reskill, they will face hardship on a personal level.

Understandably, especially painters and other visual artists have been outspoken about the use of their work as training data for AIs like Dall-E, only to be then confronted with the AI as a direct, cheaper, and faster competitor. Possibly regulation might help alleviate some of this pain, for instance by providing artists with ways to stop their work being used as training data, or potentially to be compensated for allowing AI companies to use it.
However, this will not stop these tools from becoming better, cheaper, and more accessible.

The fields where we currently see this disruption are creative. As anyone who works in a creative industry or hired a designer in the past knows, the great differentiator between better and worse creatives is in understanding the client needs and delivering the right solution. Both companies and individuals trying to address the new AI challenge are — in my view — well suited to invest in their relationships and the skills to understand clients as well as possible. This could entail communication or even systemic coaching skills, and market insights to foresee better what their clients need. A second recommendation goes to a wider audience:

Get used to AI by using it

AI will stay. If I am correct, it will even lead to industry disruptions, meaning rising competition in the low-margin areas of creative industries first, but then moving to higher-margin areas and possibly to other industries.
I like to compare situations like this to the introduction of Excel or Photoshop or the smartphone. Nowadays, it is unheard of companies and professionals not using one or many of these tools. So the same will be true for AI. In order to be ready for it, try and test it now. AI can be used both in how you work and in your products. While the latter is still a bit more difficult — building AI tools into your products requires data, skilled data scientists and IT professionals, and a good sense of where the business case makes sense — the former is easy to start: why not use ChatGPT next time you have to write a proposal? It can outline your arguments, you can let it reformulate in a different style, just be creative.
Trying to incorporate a technology which powers disruptive innovation is a good idea. Don’t expect direct business impact from it, but train your muscle now.

In the table below I collected some basic actions to take depending on industry. Of course, those are neither mutually exclusive nor exhaustive.

---