The rise of AI will only make Big Tech more powerful, NYU researchers warn

• The worst-kept secret of AI is that it's really expensive to develop. That gives Big Tech companies an advantage. 
• Even startups need to rely on cloud platforms like Amazon's or Microsoft's to build their products.
• NYU-based research group AI Now says we have to address this imbalance of power before it's too late.

AI is expensive. The data models that underpin the whole affair are intensive to develop, requiring sophisticated and powerful calculations to be made useful.

The good news is that the tools, resources, and raw computing power to build those data models are readily available to pretty much anybody — leading to the recent explosion in AI tools, from OpenAI's famed ChatGPT to Runway's text to video editor. 

The bad news, say some advocacy groups, is that the boom is just contributing to the power and influence of the relatively few companies with the infrastructure and capabilities to support those intensive AI applications. Just about every AI tool you can think of relies on clouds like Amazon Web Services, Microsoft Azure, or Google Cloud, or some combination of the three. 

In practical terms, it means that even the AI startups competing with those Big Tech giants are, in a very real way, simultaneously contributing to their respective bottom lines. And the more those startups use those major cloud platforms, the more they pay to their much larger competitors. 

This dynamic shows that the rise of AI will only make Big Tech more powerful and entrenched in several industries, even as worries about their potential for monopolistic practices peak.

Researchers at AI Now Institute, a New York University-based research organization founded in 2017 that until recently advised the Federal Trade Commission on technology matters, says that the AI boom means it's now or never to do something about it.

Last week, AI Now released a report detailing Big Tech's impact on AI development — finding that  AI development has been "foundationally reliant" on resources controlled by Big Tech, including data and computing power. AI Now said Big Tech has also been positioned as a crucial part of the US-China race, giving them geopolitical importance. 

The report said much of the narrative around AI development has been shaped by Big Tech, from the idea that AI needs unrestricted innovation for social good to connecting AI development to societal progress. AI Now believes this narrative has been beneficial for large companies and has even influenced how many policymakers think about technology. 

"If we want to make any meaningful reform in the tech industry, we need to start by confronting the concentration of power," Sarah Myers West, managing director of AI Now, told Insider. 

AI is lopsided competition between big and small players

An important piece of this dynamic, too, is that the tech titans are competing head-to-head with those startups and smaller players. The difference is that they have access to resources and data not available to outsiders. Myers West calls this notion "gatekeeper control." 

For example, Bloomberg reports that Microsoft warned customers of its Azure cloud that they cannot use data from the Bing search engine to train their AI. Meanwhile, Microsoft — and its hand-picked AI partner, ChatGPT creator OpenAI — are building new products using that very same data internally.

There's no doubt competition exists in the AI market. OpenAI moved to strengthen its first-mover advantage by releasing the latest version of its flagship GPT model just a few months after ChatGPT became viral. Google is reportedly scrambling to put out its version of an AI-powered search engine to stave off Microsoft's ChatGPT-powered Bing, too.

Myers West said there is a way to encourage competition and lessen the reliance on Big Tech. 

"If there's one thing we've learned over the past decade of pushback against Big Tech, it's that we can't let companies be the leaders in this conversation," she said. "It's for the regulators and the public to define the future of AI."

AI Now presented several policy suggestions to discourage further entrenchment of Big Tech. One of these is the concept of data minimization, or allowing companies only to collect only the necessary data and nothing more. Big Tech companies often hold the most data, and most often, they have the computing power to make use of it.

Myers West said better enforcement of antitrust laws and connecting competition with the concept of privacy could limit just how big Big Tech can get. Once these are in place, it offers others a more even playing field. 

"How we get there is through stronger competition enforcement, and the good news is that competition regulators are paying attention," Myers West said.