#31: AI + Resistance

Where do we draw the line on automation?

Last week, the WSJ published an article about Waymo’s rapid growth over the past few years, as well as society’s growing acceptance of self-driving cars. The autonomous ride-hailing company is on track to complete 20 million paid rides this year and continues to expand its presence in major US cities. It’s fascinating to see that in just two years, Waymo has convinced millions of riders to opt for technology over their human chauffeur.

Personally, I haven’t taken an autonomous ride-share trip yet. But I would give it a try. In June 2023, I spent a long weekend in San Francisco and saw Waymos cruising around the steep streets, gathering necessary data points on practice runs. Seeing this made me think, “what if driving became fully automated? Would I miss driving if I never did it again?”

Taking this a step further, AI will continuously automate all kinds of processes as time goes on. And there’s plenty of things in my life that I would be thrilled to automate (cough grocery shopping cough). But what types of processes would I find a decrease in quality of life if they were automated? What am I anti-AI on?

From what I’ve observed so far, AI has seen the widest adoption by businesses, looking to save time, money, or both when it comes to day-to-day operations. That’s why there are so many AI customer service agents or AI code editing agents that have flooded the market. For decades, companies assumed that staffing a call center full of people and spending all your funding on expensive engineers were the baseline requirements of growing a business. These quite clearly inefficient systems have been disrupted by technology that can, in a lot of cases, do a better job than humans. Or at least, save money and do the job quicker.

Crediting my own diligence and networking, I’ve gathered that business AI adoption will come before consumer AI adoption. My take is that there are fewer consequences of business AI adoption, which make it easier for businesses to take the risk. On the other hand, it changes how people behave, interact, and even view themselves.

Putting aside their business use case, ChatGPT (and other LLM chat interfaces) has drastically changed how humans search for information. But they also are trendsetters. When ChatGPT released its image generation feature, the tool went viral with people posting the images they created on social media (remember Studio Ghibli?). People felt a sense of pride for what they were able to create with a novel, new technology. In a lot of ways, ChatGPT’s image generation feature looked like magic when it was launched.

But with ChatGPT, people have been mostly excited about its efficiency and effectiveness, compared with Google’s traditional search engine. There will be AI applications where people are less interested in adopting, as AI will disrupt hobbies, and even identities.

Here are a few consumer AI use cases that will see more push back on adoption than others.

Driving

Put aside general safety concerns regarding riding in an autonomous vehicle, some people really enjoy driving. And they’re not looking to get automated any time soon.

Motor vehicles are a culture. Heck, millions of people watch racecar drivers compete each week in either an F1 grand prix or a NASCAR event. And that’s not counting all the movies that have been made. From the Fast and Furious franchise to Pixar’s Cars, vehicles are embedded in both American and global culture. People like cars.

I believe there would be a huge resistance to all cars becoming autonomous (thanks to AI breakthroughs). It would be a bummer if AI replaced the feeling of driving down a tree-lined suburban road with the windows down, blasting your favorite song in the middle of July. There’s an emotional component to driving. It’s an escape. A kind of freedom you can feel in your chest.

Where I see a likely middle ground is that all ride sharing services become autonomous. This will certainly lead to some serious employment concerns for the gig economy, but it’s likely where the act of driving is going.

I do think this point is worth calling out. As self-driving cars continue to collect more data through rides, it’s not unreasonable to think that they’ll become safer than human-operated cars. As per the WSJ article, “there are millions of car crashes in the U.S. every year”, so it’s not crazy to think that we may see a decline in car crashes with more self-driving cars on the road (the article also states that Waymo has a safety record of 50 million driverless miles).

Any number above zero is too many. But that doesn’t mean society will give up control so easily. But will the goal of a safer society overrun the ability to operate a car? Likely not any time soon.

Shopping

I don’t enjoy shopping. However, the phrase “retail therapy” exists for a reason. Some people receive a dopamine hit when they swipe their card or click “purchase”. There’s an emotional aspect of shopping. It feels good to treat yourself and plays on primal instincts to accumulate resources.

In last week’s AI + Shopping, I proposed that commerce will split into two categories: monotony and discovery. All routine, daily purchases will likely be automated and executed by personal AI shopping agents. By automating these monotonous transactions, it’ll free up time for higher-consideration, experience-driven purchases.

Still, what if you enjoy shopping, no matter the item? It’s worth mentioning, shopping empowers choice, leading to feelings of validation. The emotional connection is real.

I see shopping playing out a little differently than driving. I believe that monotonous commerce will continue to eat into discovery commerce, as we come to terms with more and more purchases can be realistically automated. But those big-time purchases, like buying a house? AI may smooth out the process, but I can’t envision a world where things that cost a high percentage of your net worth would be executed by an AI agent.

What are you hesitant to automate?