Back in 2016, in his book “The Rise and Fall of American Growth,” economist Robert Gordon wrote that the century that followed the Civil War gave rise to technological innovations that spurred leaps in productivity — for both businesses and individuals — that will never be matched.
In fact, he argued that the post-1970 technical marvels, measured in computing power and the internet, pale in comparison, at least as measured in productivity improvement, to the advances wrought by indoor plumbing, television and electricity.
Generative AI’s poised to change all that, as it turbocharges the ways in which information is harnessed to solve real-world problems — right here, right now.
Maybe not.
As Amias Gerety, partner at QED Investors, told Karen Webster, Gordon would likely be impressed by generative AI — but the technology would not sway his views on technology and productivity.
As he remarked to Webster: “Time is the scarcest of resources. And electricity generates more ‘productive time’ in a day — and more mechanically and predictably — than AI will generate, in terms of more time by making certain tasks easier.” Indoor plumbing, to use another example, increases life expectancy (by making daily life more sanitary), which increases productivity, too, and standards of living, in ways that AI won’t match.
“We’re not ready to put AI into that pantheon,” Gerety said, alongside the economic revolution spurred by, say, the advent of the car.
Generative AI has indeed made some strides in saving time that’s been traditionally allocated to certain job functions within the economy. And the immediate impact of AI can be measured in the fact that the cost of knowledge-based work — content, video and image creation — is decreasing.
The cost to generate new code, he said, is going down. The cost to explore new ideas and research what promise to be breakthroughs in technology and science and health is going down too.
“Real innovations come from a combination of insight and hard work,” he reflected, “and as the ‘perspiration side’ gets easier, the possibility of finding out whether various ideas lead to worthwhile innovations” improves too, through accelerated times to market.
Viewing technology’s evolution and the promise and reality of seismic shifts can be viewed through three platform shifts that have occurred, and are still occurring, through the past several decades.
Gerety listed them: The initial shift at the end of the last millennium was the move from desktop to mobile, then came the shift from server centers to cloud-based systems. And during those decades, we saw a few “false” shifts — voice computing has not quite made the cut. Crypto certainly has not.
Now comes generative AI, which Gerety said, has all the makings of a platform shift, driven by large language models. (Incidentally, he mused, AI can do much to make interactions with voice technology a lot easier by making voice memos as easy to consume as they are to create.)
It’s an exciting moment for VC and private equity investors, noted Gerety, who likened the process of investing in the new technology to being handed a magic potion.
There’s no way of knowing if the potion turns one into a superhero — or proves deadly.
It’s venture capital’s job, he said, to “lean into those scary pathways.”
Easier said than done. The level of interest and excitement over AI is evident in the sheer volume of pitches that cross Gerety’s desk on a daily basis — and have been coming fast and furious through the past few years — he said, where AI is mentioned in every investor deck and appended as a prefix to every new product and service.
As he told Webster, there’ll be companies that receive funding and notch high valuations because they’ve got AI in their pitch. But that’s not the same thing as finding the startups that will succeed because they’ve hitched their stars to AI.
“When I’m looking at a business plan,” he said, “I’m always looking for the startups that can credibly say that ‘because of an insight, experience or expertise I have, what’s hard for others is going to be easy for me.’” Against that backdrop, he said, the startup that thrives will be better relative to its competitors (and stand out as an investment opportunity).
At the moment, he said, the most interesting areas where AI is being applied lie in fraud and risk management and underwriting — because those are industries where practitioners are sophisticated and can use AI as tools to get fast answers. There’s a multitiered approach, he said, that comes as GPT is used for speed, but proprietary analytics and machine learning are “layered” on top to ensure that results are accurate.
In one experiment he’s conducted personally, Gerety said he recently worked with an anti-money laundering attorney to use GPT to create an anti-money laundering policy — and then translate it into Chinese. What GPT returned was useful, if far from perfect.
“GPT,” he said, “is a ‘first draft’ machine.” But there’s no real substitute for thinking deliberately, to get words and ideas in the right place, to make sure that things make sense (the last time he used ChatGPT, he said, was to help brainstorm art project ideas with his kids for school assignments).
Any discussion about AI seems to include the counter-argument to innovation, which is that we’ll see an apocalyptic scenario akin to a “Terminator” movie, where the machines rise up against us in ways no one might have expected.
And that, of course, gives rise to a discussion on regulation.
“My primary advice here is to avoid the temptation of going down the technological rabbit hole,” he said.
When technologists head to the Hill to debate policy, the mindset seems to be that the lawmakers have to get a lot of smarter, and need to understand the technology. That’s a mistake, he said. Members of Congress have to grapple with a plethora of constituents’ needs, and are often understaffed, so getting up to speed on all of the nuances of AI would be a fruitless pursuit.
The best approach is to identify simple questions tied to the intersection between tech and daily life that get to the heart of the matter. The Turing Test — named after computer scientist Alan Turing — is a yardstick here, where the focus is whether a computer can “fool” a person into thinking the computer is human — and the Turing Test should be used to help create policy.
“This is a place where I hope that technologists will be focused less on ‘Terminator’ scenarios and be focused more on what’s happening today and tomorrow.”
We’re always on the lookout for opportunities to partner with innovators and disruptors.
Learn More