Big Tech firms like Apple and Google’s parent company, Alphabet, are likely to deploy artificial intelligence to maintain industry dominance, according to Cathie Wood’s investment firm ARK Invest.
On Oct. 1, ARK released a white paper dedicated to the disruptive capabilities of AI, sharing its framework for identifying disruptive technologies.
Source: Cathie Wood
Authored by ARK’s chief futurist, Brett Winton, the report explores how Big Tech companies may strengthen their AI implementation to maintain industry dominance, while also pointing out the potential for failure of such strategies.
AI had the “steepest cost decline curve of any technology in history”
Winton outlined three core properties of disruptive technology platforms: steep cost declines, penetration into new or under-served markets and business models that delay monetization and may seem “financially unattractive.”
Disruptive technologies often enable smaller firms to compete with Big Tech companies, even when industry giants recognize the technology’s potential and try to harness it for business gain.
According to Winton, AI is “absolutely” a disruptive technology, with the “steepest cost decline curve of any technology in history.” He added:
“The cost to operate artificial intelligence models of equivalent performance has been halving every four months — a trend that we expect to persist throughout this decade.”
Winton compared AI’s cost decline to Moore’s Law in the semiconductor industry, which implies that semiconductor costs are cut in half every 18 to 24 months. The data suggests that the “AI revolution is moving four to six times faster,” Winton said.
Cost of a computer that can generate GPT-4 class output at 50 words per second. Source: ARK Invest
Big Tech tends to let startups “de-risk” tech like AI
Winton also noted that incumbent tech companies often allow startups to de-risk new technologies before adopting them at scale.
“Google and Apple have taken that approach to AI,” Winton wrote. He pointed out that Google didn’t publicly release a large language model until OpenAI had been in the market for more than three years.
Related: OpenAI’s move to for-profit: Is it indeed ‘illegal’?
“Even then, despite marketing demos that seemed to indicate otherwise, Google’s performance lagged,” Winton said, adding:
“Indeed, since early 2023, using the most advanced Google model instead of the most performant OpenAI model would have cost customers 40%+ more in per-unit performance.”
Google versus OpenAI’s price and performance difference. Source: ARK Invest
Google and Apple have good reasons for delaying AI features
While Google has trailed OpenAI, Apple has yet to launch a large language model, Winton said. Apple is expected to debut its first modern AI-driven products later in 2024.
“Slow does not mean necessarily that a competitor will lose the race,” Winton said, noting that companies like Google and Apple have good reasons for delaying the introduction of AI features. One reason is that shipping a product that performs in unpredictable ways can be “terrifying for the stewards of a carefully developed reputation.”
Still, Winton questioned whether such an approach will allow Big Tech companies to introduce AI systems that become widely performant.
“They are incumbents. They would prefer technology that is not so disruptive. To their detriment, it is,” he concluded.
Magazine: Advanced AI system is already ‘self-aware’ — ASI Alliance founder