OpenAI CEO Sam Altman is reportedly seeking up to $7 trillion in funding to reshape the global semiconductor industry to power artificial intelligence. The fact that a company can come up with a larger financing target than Japan's GDP and not be ridiculed out of the market is another sign of the high concentration of the biotech market.
From the promise of medical breakthroughs to the dangers of election interference, from the hope of beneficial climate research to the challenge of cracking fundamental physics, AI is too important to be monopolized.
However, the market is moving in this direction, as the resources and talent to develop state-of-the-art AI are firmly in the hands of a very small number of companies. This is especially true for resource-intensive data and computing power, known as "compute," that is necessary to train large language models for a variety of AI applications. Researchers and SMEs will run the risk of becoming fatally dependent on Big Tech once again, or they will miss out on the latest wave of innovation.
There is a frenzy of public investment on both sides of the Atlantic in an attempt to level the playing field in computing. In an effort to ensure that scientists have access to capabilities comparable to those of Silicon Valley giants, the United States established a national AI research resource last month. The pilot project is led by the National Academy of Science. In partnership with 10 other federal agencies and 25 civil society groups, it will promote the granting of data and computing to help the research and education community build and understand AI.
Back in 2018, before the recent wave of generative AI brought a new sense of urgency, the European Union built a decentralized supercomputer network with similar goals. Eurohpc has been relatively unknown, and the program doesn't seem to be fully utilized. As European Commission President Ursula von der Leyen said at the end of last year: we need to use this power. The EU now believes that democratized access to supercomputers could also help create "artificial intelligence factories" that allow small businesses to focus their resources on developing new cutting-edge models.
Internet access has long been discussed as a public utility because of its importance for education, employment, and access to information. However, the rule in this regard was never adopted. But with the unleashing of computing as a shared good, the United States and the European Union have shown a genuine willingness to invest in public digital infrastructure.
Even if the latest measures are seen as industrial policies in the new veneer, they are part of a long overdue effort to shape the digital market and offset the enormous influence of Big Tech in all corners of society.
These** make the right decisions by expanding access to underlying computing resources, but such investments are only the first phase and must be closely coordinated with legislative and regulatory interventions. Antitrust authorities must ensure that the largest AI companies don't become extraordinarily large. Security agencies must prevent malicious actors from accessing critical computing resources.
Non-discrimination regulators are busy dealing with the various ways in which AI applications exhibit bias and discrimination. Similarly, public AI investment complements policies aimed at preventing market monopolies from turning into knowledge monopolies. While the EU has wisely coded data access in academia in the Digital Services Act, imposing liability on platform companies, it has not explicitly included such provisions in the AI Act. For example, companies will have to report on energy use and data entry, but trade secrets will be respected, allowing for significant opacity of key details.
Going forward, investment in public digital infrastructure must increase – state funding must be diverted from big tech companies, even if they are used for projects with public functions. In 2022, the United States** invested $3.3 billion in artificial intelligence, which is a considerable amount, but it is nothing compared to the tens of billions of dollars invested by industry each year or the trillions of dollars that Altman seeks.
Preventing AI monopolies is part of a healthy climate of innovation and is increasingly important for the public to better understand the technology. In this case, the goals overlap. Historically, academic research has been at the root of many valuable innovations. This ecosystem must not be stifled.
#ai#