The debut of ChatGPT by OpenAI last year received both praise and criticism from industry analysts. It was lauded for its potential to disrupt a wide range of vocations, including computer programming, education, financial trading and analysis, graphic design, and painting. Universities hurried to adapt their curricula in anticipation of AI, rendering college essays obsolete.

Among the immediate consequences predicted by ChatGPT was the potential reinvention or even replacement of established internet search engines. This sparked speculation about whether chatbots could pose a threat to Google, given that search and related ads account for the vast majority of Google’s revenue.

While ChatGPT demonstrated impressive machine-learning skills, it was not a stand-alone service. The necessity for a partner to harness OpenAI’s technological prowess led to a quick collaboration announcement with Microsoft. This collaboration between the AI start-up and the legacy tech company poses a real challenge to Google’s dominance, heightening competitiveness in the “AI arms race.” It also provides useful insights into the aspects that will influence a company’s success or failure in using this technology.

To appreciate OpenAI’s decision to align with Bing and assess if Google can still maintain its dominance, it is critical to grasp the differences between this technology and previous breakthroughs such as telephones or market platforms such as Uber or Airbnb. In each of those situations, network effects, in which the value of a product increases with its user base, played a critical part in determining the different companies’ growth and success. ChatGPT and other generative AI services are prone to a similar but unique form of network impact. Managers and entrepreneurs must understand how these new AI network effects function in order to develop effective AI integration strategies.

AI differs from traditional products and services because it relies on accurate forecasts and suggestions. In contrast to converting supplies into outputs, AI causes enormous data sets that must be constantly updated through consumer interactions. To remain competitive, an AI operator must collect, analyze, and use data to make predictions and solicit feedback to improve suggestions. The system’s value is directly related to the data it receives from users, resulting in the formation of data network effects.

Data network effects, also known as data-driven learning, are separate from direct network effects seen in devices such as telephones or indirect network effects shown in platforms such as Etsy or Airbnb. In data network effects, the technology’s performance and value rise as the user base grows. However, the value is not determined by the number of peers or buyers and sellers. Instead, it stems from the technology itself: AI improves through reinforcement learning, in which feedback follows predictions.

As the AI system improves in intelligence, it makes more accurate predictions, making it more helpful and enticing new users while maintaining old ones. With more users, the system receives more responses, improving forecast accuracy and establishing a positive feedback loop. As a result, data network effects promote the evolution and advancement of AI technology.

The collaboration between OpenAI and Microsoft tackles the limitations of ChatGPT by exploiting the feedback loop of Bing users. The partnership allows OpenAI to update and improve ChatGPT by testing predictions and grading responses. The next stage might involve Microsoft using its massive user data from Excel, PowerPoint, Word, and LinkedIn to increase the AI’s capacity to replicate varied documents, which would assist office workers. This collaboration enables OpenAI to overcome obstacles and improve its AI technologies.

The collaboration between OpenAI and Microsoft tackles the limitations of ChatGPT by exploiting the feedback loop of Bing users. The partnership allows OpenAI to update and improve ChatGPT by testing predictions and grading responses. The next stage might involve Microsoft using its massive user data from Excel, PowerPoint, Word, and LinkedIn to increase the AI’s capacity to replicate varied documents, which would assist office workers. This collaboration enables OpenAI to overcome obstacles and improve its AI technologies.

Second, executives must formalize the diligent collection of information. We can find valuable data in a variety of locations, both within and outside the organization. Companies can improve AI forecasts by collecting consumer interactions, operational records, and external elements, such as weather. Even occasional data points beyond their control, such as recruitment buzzwords or consumer browsing behaviors, have significant worth.

Finally, everyone should recognize the significance of the data they share, if intentional. Sharing facts and criticism is essential for improving predictions, but it also allows others to recognize the worth of that data. Executives must carefully assess who will benefit from the data they share or allow access to, and in some situations, limit sharing. For example, when Uber drivers use the Waze app for guidance, they accidentally contribute vital data to Google, the owner of Waze, about the frequency and duration of ride-hailing trips.

This information could be extremely useful for Google’s autonomous taxi operations. Similarly, when companies like Adidas sell on Amazon, Amazon can assess demand and price sensitivity across brands and categories, which might assist Amazon’s private label goods or competitors. Executives can mitigate these risks by avoiding platform intermediaries, establishing data access agreements, and keeping direct communication with customers.

Collaboration via data exchanges, similar to how banks communicate creditworthiness data, can also be a viable option. Understanding AI network effects, as seen in the “OpenAI & Microsoft Challenge Google,” demonstrates early adopters might gain substantial benefits, whereas following, no matter how rapid, may struggle to catch up. Access to AI algorithms and a steady stream of data results in cumulative advantages that are tough to overcome. The entire potential of AI has yet to be realized for CEOs, entrepreneurs, policymakers, and everyone else, presenting both promising prospects and significant challenges.