Generative AI is changing the game for platforms

Artificial intelligence has long powered platforms from behind the scenes. But with the launch of technologies like ChatGPT – OpenAI’s large language model (LLM) – AI has gone from futurist fantasy to day-to-day reality. Indeed, in a poll of attendees at Platform Leaders in June 2023, almost 20% of people said they use AI tools every day, of which nearly half have played with a Generative AI model.

When new technologies take hold of the public imagination in this way, the impact is far reaching. Without a doubt, AI poses fascinating questions not only about the future of work, but also about the future role and strategy of digital platforms. Geoff Parker, Professor of Engineering at Dartmouth and a visiting scholar at MIT’s Initiative for the Digital Economy, has years of experience in networks economics and strategy, and digital platforms have long been one of his areas of specialisation.

“If you go back to the basic way that we think about platforms, we think of them as matching two types of users, often a content side, and often a consumer side, but they come in lots of different flavors in multi-sided platforms,” he explains. “The goal is to create a valuable transaction.”

These concepts of matching and value creation have become central to platform design and functionality, but Geoff sees a major shift ahead: “Data-driven matching allows for the right supply to meet the right demand that creates the most value. [But] generative AI has the potential of turning that on its head.”

Rather than needing to attract or create supply first, generative AI is able to respond to demand as it arises. This also allows for supply to be even more tailored to specific preferences, and upends the matching mechanisms that multi-sided platforms employ. Instead, value is created by on-demand generation of whatever users search for, request, or desire.

 

The AI effect on network effects

Clearly, the very premise of generative AI challenges the established function of platforms. Network effects, which are often a key element of accelerating growth and creating valuable exchanges, are significantly less necessary if the platform itself can use AI to generate what users want.

However, Geoff says, this doesn’t represent an erasure of platform processes, but rather a shift in mechanics. “You don’t necessarily need to have preexisting content to match to, but you still have to have a ton of data. These systems are trained on massive amounts of information.”

He adds: “If you think about network effects, our prior conceptualization was that when [platforms] increase in the number of users, they increase in the number of participants. Now, we’re seeing a lot of value flow to the data itself. So I think we’re going to see a shift in the locus of value creations in platforms.”

This poses fundamental questions for digital platforms about the standalone value of their products or services. A social network platform, Geoff points out, has very little value without the presence of other users. Geoff prompts platform practitioners to ask themselves, “What does [your platform] do if you’re just playing with it by yourself?” This is a first step in understanding where the value for users is being generated. “[Platforms] have to think carefully about the various sources of value that they’re providing to their users.”

Whatever changes are ahead, Geoff thinks that they are not far off: “I would expect to see the relative fraction of value start to shift around a lot. It doesn’t mean that the platforms are going away, but it means there’s an adaptation coming up.”

Emerging markets and opportunities for innovation

Along with the need for strategic adaptation, generative AI offers plenty of new opportunities as it becomes more widely adopted. One emerging market is for prompt engineers – experts in designing instructions for LLMs – but Geoff sees other possibilities. “If you buy the premise that these systems need massive amounts of data to be trained on and that creates value, [then] that would suggest either markets for data or data consortia and collaborations,” he says.

At the same time, the data sets that underpin LLMs are often proprietary. Geoff believes that this will create yet another new market for data set-up and maintenance. “Both of those are going to be forces that the platforms will have to reckon with, and they’ll create new markets.”

Overall, the potential for innovation in this field is remarkable, as Geoff illustrates with an example from his classroom at Dartmouth. “What was absolutely stunning to me – or frightening – was that within four or five weeks, students put together something that’s this close to production ready. […] With 250 lines of code they called three major systems, embedded it in two different LLMs, and [took] work that would’ve taken months and did it in weeks.”

He also notes that the education sector is uniquely positioned to fuel innovation. “Companies are super busy with their day-to-day,” says Geoff. “Students have a chance to play with this tech and can actually come in and bring some pretty new ideas and new capabilities.”

What does the future hold for platforms?

Geoff likens the mainstream arrival of AI to Mosaic, one of the first web browsers to be widely available, which was revolutionary in its integration of multimedia and text. He reminisces about his response to Mosaic when it was released 30 years ago: “I said, ‘What this technology is doing is it’s taking a lot of existing things and then integrating them to, in effect, create a new experience.’” He adds, “This is often how we see innovation working. That’s what I see here – it’s that fundamental.”

In some circles, predictions are already swirling about how generative AI will spell the end for many platforms, but Geoff isn’t convinced. “One could have said the same thing about Microsoft decades ago when they were so slow to pivot to literally embedding any kind of internet connectivity. And yet, here they are,” he says. “So I wouldn’t be so fast to write off the firms that are getting described there. What I would expect to see more is a really intentional set of pivots and then adaptations to the sources of value creation.”

As far as the role of regulation, there are three issues that Geoff predicts will be highly relevant for regulators in a digital ecosystem where AI takes on a larger role: data ownership and access, deep fakes, and incorrect information. For all that these areas pose steep challenges, however, Geoff doesn’t think that the drawbacks outweigh the benefits. “Both the firms themselves and the regulatory community have to work to understand the issues, and then don’t let the perfect be the enemy of the good,” he thinks. “[It’s about] trying things to blunt the malign forces while still harnessing what’s good.”

AI: disrupting or augmenting the future of platforms?

Geoff acknowledges wryly that many people might try to frame AI as being either a destructive disruption or filled with augmentative potential, but the reality is more complicated and contextual. In an environment where data can be leveraged to generate exactly what users want, Geoff thinks that some platforms stand to see major augmentation of their business: “If you think about who has the best data, it’s the big platforms.” However, realising this potential depends on how platforms react. “I think it’s very disruptive for those firms that can’t pivot their business model or can’t pivot how they think about their internal operations and their ecosystems and where value comes from,” he says. 

“I think we’ll continue to see [platforms] make progress, particularly into traditional businesses,” he continues, citing B2B and the manufacturing supply chain as two examples. Ultimately, in Geoff’s view, there is also no escaping the rate of change. “The timeline for having to ingest an understanding and these tools into your organizations, and [adapting] the way you create value, is just going to increase,” he says. “I think nobody gets to ignore these forces.” 

“The future of platforms is…always adapting. It’s unstable — but it never has been stable. – Prof. Geoff Parker”

To go further

This keynote with Prof.Geoff Parker and Laure Claire Reillier was part of the Platform Leaders event organised by Launchworks & Co on the 7th of June 2023 (full list of speakers and agenda). To watch the full event, play the video below.