Investment and competition are escalating as the research & development labs of startups and tech giants race to shape the future of artificial intelligence.
Boyden's perspectives on the news and trends that are transforming industries
ChatGPT rebooted the conversation around artificial intelligence when it launched in November. Within five days, the AI chatbot from OpenAI attracted one million users. Microsoft invested $10 billion in the company, building on a $1 billion investment in 2019. Other tech giants are eager to roll out rival chatbots. On February 6, Google launched Bard. Days later, Microsoft launched a new AI-powered Bing search engine. Chinese search giant Baidu is on the verge of releasing an AI chatbot of its own.
ChatGPT and the surge in generative AI that it sparked in 2022 mark a milestone in the development of artificial intelligence. It remains to be seen whether the technology will change the world. But it has certainly ratcheted up the amount of energy and capital flowing into AI. It is also changing how the tech industry manages innovation, particularly corporate research labs, which combine the processing power of Big Tech and the talents of top computer scientists.
Countless scientific advances have emerged from corporate R&D, particularly in post-WWII America. Companies invested heavily in scientific research to develop useful products – and they succeeded, making corporates like Dupont, IBM and Xerox household names. However, by the late 20th century, R&D increasingly prioritized development over the basic science carried out in research labs. This meant fewer new ideas, and more applications for existing ones. Science was largely relegated to universities.
Now, with AI taking great leaps forward, the R&D dynamic is changing again. Agile startups are entering the scene, but according to The Economist, most of the recent breakthroughs in artificial intelligence are coming from tech giants. They have the computing power, and can apply the results of research to commercial products more rapidly. Amazon produces two-thirds as much AI research as Stanford University, America’s top school for computer science, and Meta produces four-fifths as much. Alphabet and Microsoft generate far more, and are now the top two contenders in generative AI.
For the moment no single AI model dominates. David Ha, formerly a research scientist at Google and currently Head of Strategy at Stability AI, a consortium of small firms, universities and non-profits, believes this is because AI knowledge is quickly dispersed among cohorts. Researchers from competing labs “all hang out with each other”, he says. They also tend to move between organizations, and being scientists, they are keenly interested in publishing and presenting their research.
Corporations do not necessarily share this openness. Competition is heating up, particularly between Microsoft, which has OpenAI, and Google, which has DeepMind. Being a small firm, OpenAI may have more freedom to release products. This would enable it to collect more data, and use it to improve its models. But so far in the development of generative AI, bigger has been better. Tech giants have the advantage here. However, technological development could take another direction. There are limits to how big models can get, and more specialised AI models could come to define the standard.