AI revolution: Artificial general intelligence is exciting
Its potential, for good and for bad, is just starting to be appreciated.
[Editor’s Note: Today is the first of two artificial intelligence (AI) pieces that Linda did before going on vacay. Part 2 will be published next week.]
Consider this. On average, every minute of every day sees 2.43 million Snapchat snaps, 16 million cell phone texts, 231.4 million emails sent, 5.9 million Google searches, 1.7 million new pieces of shared Facebook content, 1.1 million Tinder swipes, 347k “tweets’’ (soon to be Xs? How do I X?), 66k shared Instagrams, $100 million worth of purchases using crypto, $433k spent on Amazon and $438k of Venmo user transactions. What does this have to do with AI? Everything. The virtual world has become the real world, with new technologies intertwining with the way we work and live at an accelerating pace. According to Citi Research, it took seven years for the World Wide Web to reach 100 million users globally. Five years for Twitter. Two for Apple’s App Store. Nine months for TikTok. Launched on Nov. 30, 2022, ChatGPT reached 100 million users in just two months.
AI is not new—"Siri,” online chats, Grammarly, telephone trees, even internet searches are some common ways it already has woven into our lives. But ChatGPT is the poster child for a new era of generative AI, an umbrella term that encompasses a range of its productivity-enhancing technologies. Machine learning (using data and algorithms to imitate the way humans learn). Cognitive computing (computerized models that seek to simulate the human thought process in complex situations). Neural networks (processing data in a way that is inspired by the human brain, such as the large language model that gives ChatGPT the ability to generate coherent and contextually appropriate content). All allow tasks to be conducted and skills to be improved without human coders or enablers. They learn and refine on their own, their DNA the applications, coding, algorithms and parameters provided by their creators.
Generative AI’s potential is just starting to be appreciated by corporate America. More than third of S&P 500 companies talked about it in Q1 earnings calls, and Bank of America expects nearly a third will incorporate some form of it into their processes by 2025. Design optimization, risk analyses, materials use, predictive maintenance and automation are just some examples. Jefferies estimates the subsequent cost savings could range from 21% to 26%. Lower skill and repetitive jobs that are easy to replicate may be most exposed to AI disruption. This already is going on as warehouses, factories and fast-food outlets have been turning to robots, automation and self-serve/self-checkout at an accelerating rate—McKinsey estimates 11.8 million workers in office support, food and customer service, and production may need to find a new kind of job by 2030. What many may not appreciate is AI also coming for higher-skilled, higher-paid white-collar jobs in such areas as communications and media, capital markets, insurance and banking, even software and platforms. Overall, a McKinsey study of 63 practical uses of generative AI across 2,100 existing jobs found activities that consume about two-thirds of our work time today likely will be automated over the next 20 years!
This doesn’t mean workers heading to unemployment lines. While Goldman Sachs estimates 18% of work globally could be automated by AI, and nearly 25% in the U.S., tech innovation historically has led to the creation of new occupations that have accounted for the bulk of employment growth over the last 80 years. Moreover, there’s always churn in the labor market—from 2019 through 2022, 8.6 million workers changed occupations. “Do not fall into the Luddite trap of thinking radical productivity-enhancing technology kills jobs. History proves just the opposite,’’ says Piper Sandler. In fact, AI integration often enhances, not replaces work. Productivity at one software company rose 14% as AI helped new hires mimic the behavior of experienced sales reps. By stimulating workers’ desires for “human only” interaction that can deepen their sense of belonging to a real vs. virtual community, AI could even mean sayonara for WFH. A Deloitte study sees increased spending on corporate retreats, social events and such as key to combatting AI fatigue and found companies that already do this tend to outperform peers.
The benefit from AI for companies: higher productivity and fatter margins! Goldman Sachs estimates widespread adoption of AI could lift compound annual S&P EPS growth to 5.5% over the next 20 years vs. 4.9% its model currently assumes. The big unknown is the timing and ability of companies to achieve this incremental lift. It took 10 years from their introduction before cell phones saw their biggest unit growth—14% compounded annually from 2003-13, Jefferies says. AI, of course, isn’t starting from scratch. The broader benefit may be to society. Declines in working-age populations in developed economies and China are projected to deepen through 2050, and annualized productivity in the U.S. is running well below its 1945-65 average. Goldman Sachs believes AI potentially could raise global productivity by a percentage point annually over a 10-year period, and an even larger 1.5% annually in the U.S. Brookings Institution thinks the possible boost in the U.S. could be an even higher 1.7 percentage points, enough to lift us back to our post-World War II heyday. Bank of America/PwC projections show GDP rising across the globe and industries due to AI, with the biggest gains in North America and China by regions and health care, education and other services by industries.
To me, the mood feels reminiscent of the late ’90s and the original dot-com boom. By 2000, we had companies trading at P/E multiples in the thousands—or not at all since often they were still losing money and, in some cases, not even revenues. There’s a similar buzz about AI in conference calls, and the same bidding up of select stocks as the market creates its own momentum. The difference this time is many AI-influenced stocks have very solid, very visible cash flows and with a few exceptions, are nowhere near the eye-popping valuations of the dot-com darlings at previous bubble peaks. In fact, some of the biggest names in Big Tech also are the biggest players in emerging generative AI. Unlike Pet.com, they don’t need cute commercials to attract customers and grow their businesses.
There are possible headwinds. Maybe a slower-than-expected uptake of AI by companies. Promised improvements in productivity are great, but AI implementation can be difficult to get right. Employees themselves may oppose the adoption, especially if they see AI as a replacement, not an enhancement. And we haven’t even talked about potential regulation. But I’m a glass half-full girl. What’s fascinating is how quickly this new AI era is taking place. We’re keeping an eye out for opportunities, for adopters and for creators. And, as we’ll discuss next week, a brave if not frightening new world. Have you seen “2001: A Space Odyssey” and Hal? “I’m sorry Dave. I’m afraid I can’t do that.” Still gives me chills!
- Kitchens A smart fridge suggests a recipe based on ingredients the user picks, sends the recipe to a smart oven, which sets the temperature automatically and notifies the user when dinner’s ready.
- Health care Deloitte projects spending on health care to double from $4.1 trillion in 2021 to $8.3 trillion in 2040, with AI elevating proactive care from a third to nearly half of expenditures.
- Autos It’s estimated fully autonomous vehicles will generate 4 terabytes (4k gigabytes) of data per car/truck daily from their reliance on sonar, radar, lidar, cameras and GPS.
- Agriculture AI can slash a farmer’s production cost structure by up to 40%, Bernstein says, through efficiencies in the areas of fertilizer, seed, labor and chemicals.
- Banks and retail A McKinsey report says these industries likely will be first in line for the biggest boost from generative AI, with 75% of the productivity gains coming from just four business functions: customer operations, marketing and sales, software engineering and R&D.