ChatGPT, an “intelligent” chatbot, represents a major breakthrough. One would have expected that Google, the leader in search engines for the last twenty years, which has been investing heavily in artificial intelligence, would have been at the origin of it, but it is not the case. Is Google the new victim of the innovator’s dilemma, a syndrome often observed when a leader is overtaken by a new entrant?
When asked by employees about the “missed opportunity” of launching a competitor to ChatGPT, CEO Sundar Pitchai defended himself by explaining that a conversational bot is a significant reputational risk for a large company, adding that Google should be managed more carefully than a start-up like OpenAI (the originator of ChatGPT). Pitchai’s response seems very reasonable; he is acting responsibly to protect his company, aware of the risks of rushing into a new and unproven technology. Yet, it is characteristic of the Innovator’s Dilemma, a syndrome described by researcher Clayton Christensen in his research more than 20 years ago.
The innovator’s dilemma
The innovator’s dilemma starts from a simple observation: despite their often considerable resources, leading companies tend to fail when faced with a disruptive innovation. This is rarely due to a lack of awareness of the disruption underway, a lack of resources or skills. But then what is it? Christensen explains that when a disruption occurs, the incumbent is faced with the following dilemma: if it bets on the disruption, it risks compromising its legacy business, without being certain of succeeding. It may therefore find itself with a massive, immediate and certain loss, all for a limited gain, on a distant horizon and not at all certain. But if it refuses to bet on the disruption to protect the legacy business, it risks missing the opportunity and going under. In my book A Manager’s Guide to Disruptive Innovation, I show that this dilemma explains Kodak’s failure. Perfectly aware of the emergence of digital, the company nevertheless sought to preserve its film business as long as possible, which was rational, before betting everything, but far too late, on digital. Marconi, an electronics technology company that was very advanced in the 90s, offers the opposite example: pushed by analysts, it abandoned its legacy business to bet everything on the Internet, just before the bubble burst in 2000. It disappeared soon after.
Google might be in the same situation. The company, created in 1998, earns money mainly with its search engine and associated advertising. It has launched many innovation initiatives, but most of them have failed. ChatGPT is disruptive for Google Search because one can imagine people abandoning Google Search and using only ChatGPT. After all, why use a “dumb” search engine when you can have a much more sophisticated tool? So there is a real threat of substitution. Of course, we have to qualify this threat because substitution is never complete (the VCR did not kill the movie theater). Google search is simpler, faster, less expensive in terms of machine time and free, which will probably not remain the case for ChatGPT. Nevertheless, this fear of cannibalization light explain it is not Google, which for years has been the search specialist and has invested billions in AI, that released ChatGPT. This is very similar to what happened to Sony: from the 80’s on, the Japanese company invested billions to marry content (music, movies) to the container (electronics); however, it was Apple that won the MP3 player market. Why? Because Sony was afraid that MP3 would develop piracy and compromise its CD sales.
Although Google has all the cards in hand to release an equivalent of ChatGPT and go all out, it is still a victim of the innovator’s dilemma: on the one hand, there is the fear that GoogleGPT (let’s call it that) will cannibalize Google Search and thus lower advertising revenues. On the other hand, ChatGPT, as impressive as it is, still has many limitations. This is typically the case of a disruptive technology: it offers an attractive service, but suffers from many weaknesses. Its level of performance remains insufficient for a long time for the incumbent. The first cars broke down all the time and nobody “serious” used them. A “serious” (large) company will therefore wait until the technology reach a certain level of performance in its eyes, before investing in it; In the meantime, it leaves the place to others.
Finally, as Pitchai points out, GoogleGPT would represent a reputational risk. One may remember Tay, the conversational robot, ancestor of ChatGPT, launched by Microsoft in 2016. In a few hours, it began to emit racist and misogynistic messages. It was hastily removed and caused considerable harm to Microsoft. After this episode, we understand the caution of Google’s CEO who does not want to relive the same thing. This caution, which is quite rational and defensible, nevertheless leaves the field open to competitors who do not run the same risk, especially startups. Leaving the field open to competitors with a disruptive technology is therefore not due to blindness or incompetence. It is the result of a rational reaction of the incumbent player who has to manage its risk.
The advantages of new entrants
Against an incumbent like Google, startups have therefore three advantages: they have no historical activity to protect; they can already prosper with the new technology, even if its performance is inferior; and they run a much lower reputational risk. You can tolerate things from a startup that you wouldn’t forgive from a very large company like Google. These three reasons explain why startups will always have an advantage over incumbents, even powerful ones, in disruptive situations.
The innovator’s dilemma affects all companies, including the most cutting-edge ones. The interesting thing about Google is that we are not dealing with an “old economy” behemoth, but with a Silicon Valley company. To paraphrase Karl Marx, yesterday’s disruptors are being disrupted in turn. History is repeating itself. What is surprising here is that the Dilemma has been described for more than twenty years, and that seasoned executives like those of Google do not seem to be aware that it is working against them right now. One can only recommend them, and leaders of other industries, to take a strong interest in it if they don’t want to end up like Kodak.
Note: Things move fast: Not surprisingly, Google has just announced its own solution “Bard” as a response to ChatGPT. I will write about incumbent response to a disruption in my next article.
🇫🇷Article in French here.
➕To read more on how to assess the potential of Chat GPT, read: Assessing the Potential of ChatGPT: Lessons from the History of Innovation. On the Innovator’s Dilemma, see my previous article: Disruptive Innovation: What We Owe to Clayton Christensen.