In an interview with CNBC, Todd McKinnon, CEO of cybersecurity company Okta, said that Microsoft's decision to outsource all of its AI research and development to OpenAI is good news for Google.
MacKinnon said Google actually pioneered much of the research that led to today's generative AI, and that Microsoft's reliance on OpenAI could ultimately reduce it to little more than a consulting firm.
“This all came out of Google and DeepMind and research,” McKinnon said. “So the breakthrough was the Google research, the Transformer, which is the algorithm that all these LLMs use. [large language models] We are using it to make these big advancements happen.”
READ ALSO | Laid-off Indian-origin engineers told they'll be replaced by Indians in India: 'They're taking our jobs'
Demis Hassabis, CEO of Google's DeepMind, has suggested that Google will overtake Microsoft in AI investments. According to a report by PC-Tablet, at a recent conference, Hassabis highlighted Google's commitment to investing more than $100 billion that Microsoft and OpenAI plan to spend on AI initiatives over the next few years.
According to a CNBC report, many of Microsoft's flagship products are beginning to incorporate OpenAI technology, including its generative AI chatbot, CoPilot, and PCs with generative AI software.
Microsoft has invested a total of about $13 billion (Rs 1.8 trillion) in OpenAI. In January 2023, the tech giant said the investment would “accelerate breakthroughs in AI and ensure that its benefits are shared broadly around the world.”
Still, Google has some hurdles to overcome before it can achieve commercial success with its AI investments.
Microsoft's partnership with OpenAI has effectively made it a leader in advancing underlying AI models, raising concerns that Google's position in search could be eroded as internet users increasingly turn to ChatGPT and other AI chatbots for their search needs, the report said.
READ ALSO | India's EU-like antitrust laws unnerve tech companies
Meanwhile, Google's own AI efforts have been plagued by a number of highly public missteps.
Google's AI Summary feature, which aims to summarize information on the internet when users search, has attracted attention for generating nonsensical and dangerous responses to queries, such as telling some users to put glue on pizza or eat stones. Earlier this year, Google's Gemini AI model was criticized for generating “historically inaccurate” images, according to a Quartz report.
MacKinnon noted that the development of AI works differently than other technologies: “Unlike other generations of technology, with personal computers the biggest companies in the world didn't necessarily dominate because personal computers were almost like toys and they were truly disruptive,” MacKinnon said.
READ ALSO | Sundar Pichai turns 52: The tech billionaire's journey from Madurai to Google CEO
“There are no new AI models that are toys. The only reason OpenAI was able to make it work is because they did the great R&D they needed – they got $10 billion from Microsoft to run the models – and it was a $10 billion investment, not a disruptive one,” he added.
McKinnon sees the “biggest risk” for the cybersecurity industry going forward is that AI issues caused by digital giants, such as disinformation, could “stifle technological progress”.
“potential [for] “Artificial intelligence is incredibly important,” he said, but added, “I expect we'll see significant changes in regulation, ensuring that only the biggest and most powerful companies have control over AI.”
