A sampling of recent pronouncements in the news:

• AI is going to discover miracle cures for every disease
• AI is going to cause mass unemployment and eliminate whole professions
• AI is going to make everyone wealthy
• AI is going to crash the economy

How can we possibly make sense of these conflicting predictions? Given this range of possible outcomes, how can you, and your museum, navigate the promise and perils of artificial intelligence?

I’m working on a series of posts I hope will help, starting with a set of three scenarios illustrating the range of possible futures we face:

• AI-Led Abundance: combining the brightest projections for how AI might change our lives.
• The AI-pocalypse: a cautionary tale that may motivate us to move in a different direction.
• The Muddled Middle: describing where we will end up if we proceed along our current path.

These scenarios will try to collapse the mass of news you are reading into a manageable set of possibilities, highlight the challenges and opportunities we face, and the critical choices we must make. Later in the series, I’ll tie them into suggestions for strategic planning.

First up, an optimistic glimpse at one possible future.

AI-Led Abundance

8 am: Jarvis, your personal AI assistant, sends a gentle ping reminding you that this is the optimal time for you to get up, along with a summary of recommended activities for the day. “You’ve been feeling a little down!” J. chirps, “why don’t you spend the morning at the art museum? That always lowers your blood pressure and raises your serotonin levels!”

8:10 am: the smart toilet sends an alert to your doctor, noting your glucose levels are a little high, but your stool volume and consistency is returning to normal after your bout with intestinal flu.

9:00 am: your kids start a science project in the living room, supervised by their Melania Robot Teacher. Mel, as they call it, has also scheduled a gustatory social interaction (aka lunch) with their friends next door at noon.

9:45 am: Jarvis calls a self-driving car to take you to the museum

10:00 am: as you walk through the entrance, the museum’s integrated video system recognizes you and starts adding data on your visit (path, dwell time, emotional reactions) to the Member Management System. 

11:20 am: you duck out to the museum’s courtyard to take a call from your mom who is confused, yet again, about why you aren’t “at work” in the middle of the day. “We scraped and saved and went into debt to send you to art school and now you aren’t even using your MFA?” “Mom,” you patiently explain for the umpteenth time, “I don’t have to work now, remember? Nobody does. My stipend from the AI Sovereign Wealth fund paid off my student debt and covers all my expenses.”  

1 pm: you meet your spouse at their doctor’s office for an actual, in person consult. You both feel incredibly lucky—five years ago the tumor probably would have gone undetected, and when it did surface, would have been untreatable. Now the protocol designed by the AI assisted radiomics team has successfully put the cancer into remission.

4 pm: you and spouse join friends downtown for drinks. (The tireless Mel is now overseeing the kids at a neighborhood soccer camp.) Your favorite restaurant is in a neighborhood that used to be sketchy. Now you feel safe walking the streets, even late at night, due to the deployment of ubiquitous video surveillance via public and private security cameras, and law enforcement AI that predicts and prevents crime.

8 pm: reclining in a lawn chair, hugging your seven-year-old in your lap, you point to a bright spot moving across the sky. “See that? It’s one of the Musk Orbital Data Centers that run Melania.” “Why is it way up there?” Xe asks. “They used to be down on earth, but they are noisy and dirty and suck up lots of power and water.”

10 pm: you take out your phone to spend some time online. You and your friends swore off social media for several years, repelled by the swamp of hate speech, trolling, and disinformation. Now that social platforms are required to crack down on content and bad behavior, you are beginning to reengage.

First, think about downsides

If you’ve read Ursula K. LeGuin’s iconic story The Ones Who Walked Away from Omelas, you appreciate the importance of checking the basement (as it were) of any seeming utopia. In the case of the projections above, lifted from scientific literature, popular press, and promotional material from commercial companies, you might ask yourself how you feel about:

• Trading privacy for the promise of better health and security.
• Trusting the AI algorithms underlying these systems to be equitable, and free of bias.
• Not having to work (even if you wanted to) in a future where AI takes over most jobs, but you share in the profits.
• Basing public education on personal AI tutors for children.

If that all sounds good, maybe this is a future you want to work towards. If not, you might want to start framing an alternative vision, where different values and constraints shape our use of AI.

Then, do a reality check

We can test the plausibility of any scenario by looking for hints of this future in the present day. In fact, many of the plot elements in AI-Led Abundance already exist, at least in seed form.

• There is growing evidence that AI can make significant improvements to the detection, diagnosis, and treatment of disease.
• There are already smart toilets that can analyze your poo and pee to flag potential medical issues.
• AI agents are stepping up to manage people’s schedules.
• Self-driving cars are working the streets in L.A., San Francisco, Phoenix, Austin, Atlanta, Miami, and Nashville.  
• Museums are experimenting with the use of AI for tracking and analyzing visitor behavior and reactions.

How about the financial and regulatory underpinnings of this story? Is it plausible to think we could create a world in which we’ve curbed the worst behaviors enabled by AI, and shared the economic benefits?  OpenAI, one of the major players in the race to develop and monetize this technology, recently tackled these questions in  Industrial policy for the Intelligence Age.  To create a safe, prosperous and equitable future powered by AI, the report’s authors envision a future in which the US:

• Protects labor rights by giving workers a formal role in determining how AI is implemented.
• Implements a tax system that redistributes profits generated by AI to displaced workers and to federal social safety net programs.
• Creates a Public Wealth Fund that provides every citizen with a share of the revenues from AI.
• Implements universal health care.
• Creates training pathways into “human centered work” (e.g., childcare, eldercare, education, healthcare) and incentivizes employers to improve pay and working conditions in these fields.
• Creates corporate governance structures for AI companies that embed public-interest accountability into decision-making.

(This is a selection from a much longer list of recommendations.)

Are the report’ recommendations feasible?

This future has a few toeholds in the present:

• There is a law protecting workers from being displaced by AI. (In China).
• Bernie Sanders just introduced a bill in Congress to establish an AI Sovereign Wealth Fund (The bill is too new to appear on GovTrack.us, but that site gives Sanders’ other AI related legislative proposal—a moratorium on data centers—a 2% chance of passing.)
• There is a governance structure—the Social Benefit Corporation—that makes companies legally accountable for achieving public good. (That structure is used by less than 0.05% of companies in the US.)

Could they happen fast enough?

These recommendations are designed to address disruptions that are happening now and in the next few years. How long do you think it would take to get from our current reality to the reality OpenAI describes?

What’s Next?

I encourage you to:

• Take this scenario and make it your own—are there optimistic elements you would add to the mix? Are there elements you think are dystopic, that you might delete and replace?

• As you read AI related stories in the news, think about whether they signal that we are moving towards, or away from this positive vision of an AI future.

• Consider whether the ways you use AI support an optimistic outcome in the long-term.

Stay tuned for a trip to the opposite side of the Cone of Plausibility, when I share a scenario of the AI-pocalypse.

Yours from the AI future,

Elizabeth Merritt, Vice President Strategic Foresight and Founding Director, Center for the Future of Museums, American Alliance of Museums