Milan Mirick is Assistant Professor of Data and Operations Science at the Marshall School of Business at the University of Southern California.
Amid high interest rates, the palette of venture capital has dwindled, except when it comes to artificial intelligence. Investors have poured more than $40 billion into AI companies in the first half of 2023. More than a third of the companies in the latest batch of the prestigious Y Combinator accelerator are AI startups. Almost everyone seems to be riding the AI wave. There is a lot of fear of leaving.
The explosion of AI startups in dozens of fields masks something that many of them share: they are increasingly based on the same basic technology. In the past, companies typically had to design and build their own systems. Today, most AI-based startups are based on standardized technology from a few AI companies like OpenAI, Google, and Meta.
From video games to the Apple App Store, I've consistently found that when everyone builds with the same toolbox, a slightly bigger hammer doesn't make much of a difference. Vision and strategy, not technology, help a company stand out.
Based on the increasingly standardized nature of technology we can expect three trends:
1. Companies are successful on their business models, not on patented technology.
Platform models provide common access to powerful resources, enabling individual developers or upstarts to compete with large corporations. For example, developers in the Apple App Store ecosystem don't need to build a product from scratch. Instead, they take advantage of select tools provided by Apple, reducing their responsibilities to a small part of what would otherwise be a complex technical process and paving the way for the wide range of applications we see today.
The challenge for any organization developing a set of standardized tools is making those same tools accessible to others. If you have an interesting idea, it makes it easier for others to copy it, at least on a technical level. Most new features can be replicated. If an app breaks through and finds a competitive advantage, it's often short-lived. My research has found that companies that maintain a long-term competitive advantage have better strategies, not necessarily a unique product.
Generative AI, often powered by large language models, or LLMs, are systems that can generate text, images, video, audio, code, and other media in response to stimuli. Generative AI startups need to think deeply about how to create value beyond the product's core functions through strong, recognizable branding or an optimized customer service experience. While the race to hire technical AI talent is important, building strategic human infrastructure cannot be ignored. Other employees, from product managers to marketing specialists, can find the secret sauce to stand out in a standardized market.
Standardized technology means AI companies like OpenAI, Google and Meta have more bargaining power. If an entire business is built on another company's product, the supplier may impose stricter regulations. Think about how Apple takes 30% of every app purchase and prohibits apps from leading people to buy outside of the App Store ecosystem. As with other platforms, benchmarking is very useful for providers of tools.
2. Conventional products become more uniform.
There was a time when video games had to be reinvented. He had to invent or recreate the most basic elements on his own. However, in the late 1990s, Epic Games' Unreal Engine was released, followed by the Unity Engine in 2005. Reduced production timelines and costs by allowing developers to run, animate, and build simulations.
Developers who adopted these standard game creation tools saved resources and achieved higher sales because they were able to focus on improving their products instead of building basic functionality. However, their new products were less innovative and resembled others using the same tools, suggesting potential trade-offs between performance and creativity.
The same can happen with generative AI. Products built with large language models, such as ChatGPT, are driven toward the same goal by common LLM data and processes.
At the same time, when new technology becomes cheaper, many companies can experiment with risky ideas. Non-entrepreneurs dare to try something new. For example, the Shopify store made it easier than ever to sell products online by taking care of delivery and distribution, allowing sellers to easily enter the market.
Niche AI applications rarely break out and succeed, and small businesses are in a better position to explore these ideas now that standard technology has lowered the costs and barriers to entry. However, standardized technology can make conventional AI products look the same.
3. Companies will see higher turnover rates for employees who develop or manage AI products.
My research has found that job mobility increases after the spread of a standard tool. Other scholars have argued that as spreadsheet software such as Microsoft Excel became the norm for many accounting functions, the value of skills such as arithmetic decreased and the value of spreadsheet fluency increased. Easy jobs because their skills were in high demand. We see a similar effect, albeit smaller, in programs like Salesforce, which is now so widely used that the company issues certificates to workers to tell employers how to handle it.
If most AI products are built on the same foundation, employees won't have to learn new programs from scratch when changing jobs, giving them more agency in the job market. Rather than building AI systems from scratch, those who can effectively use GPT (an AI framework developed by Microsoft-backed company OpenAI) or Google's Bard may be tomorrow's most in-demand workforce.
How to retain employees and how companies must strategize to increase turnover among workers capable of complementing AI. Standard tools reduce the investment companies have to keep employees up to speed. However, they can increase the resources devoted to recruitment and retention.
As AI products increasingly rely on standardized tools, strategy is more important than proprietary technology. Companies using these tools need to think about how they're going to create value beyond the technical features they offer, and what they're going to do to stand out from the rest.