Tired of the AI hype yet? It’s OK, I understand. I’m tired of it too. The pace of human progress — and our insatiable need to be entertained by the shock of the new — seems to be forevermore in the “hockey stick” part of the growth curve. But it’s important to remember that where you see smoke, almost always something is smoldering under the surface, even if it isn’t yet fire. So let’s separate AI fact from fiction. Here’s what I believe to be true about how AI impacts the real world, right now.
It’s easy to be distracted by all the silly stories of dumb lawyers trying to use ChatGPT like Lexis/Nexis. The real power of ChatGPT in particular is its ability to summarize, compare and contrast bodies of text.
Want a 500-word summary of the European AI Act? Easy. That’s available in GPT4 or Google’s Bard right now, with verifiable sourcing, thanks to both platforms being tied to the actual live internet.
On our webinar, Adam will walk through the development work Resolve has undertaken to leverage the existing, live tools and completely reimagine how an asset manager can connect to their own body of work. He’ll step through applications ranging from compliance documents to performance reporting; from podcast summaries to blog posts.
Truth #2: Most “AI” for consumers is meh — for now.
The easy/free versions of most AI tools, like ChatGPT 3.5, Bing’s AI search tool, and Midjourney, aren’t particularly useful or compelling. Most ChatGPT applications need something else under the hood, be it a live internet connection or a connection to a known source of truth. Even now, that’s a solved problem: Either you have to pay for higher-tier products or need to spend time head-down in Python code.
But the pace of change here is relentless. Just a few weeks ago, paying $20/month for GPT4 got you 25 questions every three hours. Now the same subscription gets you access to hundreds of plugins, connecting you to data and services from academic scientific research (Wolfram Research) to travel planning (Kayak) — even to financial data (Daizy, an investment analysis chatbot).
Truth #3: In AI land, it’s 1995, not 2001.
The comparisons between the AI hype frenzy and the Dotcom boom are obvious and endless. They’re also largely correct. We’re at the beginning of the story, not the end.
In the early dotcom era, first adopters weren’t reinventing “the Internet” every week. We were learning all the cool (and sometimes awful) things we could do with it before it became co-opted by corporations. We’re in exactly that spot now.
Trying to pick “the big winners” in 1995 as an investor was neither obvious nor easy. (I know, because I tried. And mostly I failed). But learning how the internet actually worked and following that development turned out to be a fantastic investment in human capital. That’s where we are now with AI.
Truth #4: There’s no easy AI money here.
Let’s keep a little perspective: The entire non-defense spending on research and development of any kind by the U.S. government is roughly $100 billion a year.
In 2022, OpenAI reportedly lost (invested) $540 million in developing ChatGPT. Currently, they’re looking to raise $100 billion in new funding to continue development. IDC reports they see 2023 total spending crossing $500 billion. Or 5 times the entire R&D budget for the U.S. government.
A decent portion of that spending will likely be wasted. Why? Companies will engage in an arms race from which will emerge only a handful of winners. But much of it will happen inside companies that aren’t building AI systems themselves, but leveraging those built by others to improve their otherwise non-AI businesses.
On Wednesday’s webcast, Zeno will share more insights about the specific industries currently transforming due to AI and automation.
Truth #5: (Some) AI regulation is probably a good thing.
The European AI Act is on the brink of approval, and while I’m not exactly confident in the alacrity of bipartisan governance here at home, some form of regulation likely will arise in the U.S. as well.
Having spent my entire career in the regulatory mines, I feel confident in saying that “no regulation” is rarely the correct answer to a new and exciting player in the field of capitalism, but that “misregulation” is usually worse (*cough* crypto *cough*).
Yet there are some principles I think we can and should get behind, including:
Having a fundamental right to know whether you’re interacting with humans or machines
Labeling content provenance
Prohibitions against certain predatory AI behaviors (in the EU this would include social scoring or behavior manipulation)
Basic guidelines about what “publicly available” AI must do to be considered a safe consumer product (like we do with, say, food)
Importantly, none of these principles would prevent the extreme doomsday scenarios that make for hand-wringing headlines, where “Regulate this before it destroys everything!” seems to be the usual demand. Most of those scenarios involve human beings making particularly stupid decisions (like hooking up the electric grid to an unstoppable AI) or bad actors doing stuff they’d ignore rules about anyway (North Korea deciding to hook up their nuclear codes to an AI).
But an AI bill of rights, as I’ve outlined above? That seems like a fairly basic starting point.
The AI Future is Yours. If You Want It.
Even with the title “Financial Futurist,” my crystal ball isn’t all that much shinier than anyone else’s. But what I can predict with 100% certainty is that we’re not just going to put AI back in a bottle. Nor will it end up as a narrow niche tech like, perhaps, VR helmets or Segway scooters.
The closest parallel to the experience of actually working AI into my workflow I can come up with is the day I first used a spreadsheet in the Reagan Administration. The headlines at the time were about how many office workers would get fired, and how innumerate we’d all become because “the machine” would do all the numbers work.
Yet, with our Excel templates in hand, here we still are.
A message from Advisor Perspectives and VettaFi: To learn more on this and other topics, check out our full schedule of upcoming CE-approved virtual events.
