The future of AI: As humble as Claude and safe as Waymo?

Since the advent of ChatGPT, interest in artificial intelligence has exploded. While many Americans remain in awe of its rapid development and promise, the internet is awash with concern and skepticism. People fear that it could replace human workers and poses some theoretical, but unproven, dangers.

This is not an unusual response to new technological innovation. Change tends to freak people out, even if that change is poised to make the future better. While I cannot predict AI’s future, I believe that it will be transformative, but instead of pessimism, some recent AI experiences should inspire optimism and maybe a little comedic relief.

To quote George Strait, “I still feel 25 most of the time,” except maybe when it comes to new technology. I admit that I am not an early adopter of it, nor am I particularly adept at using it. However, I have been giving generative artificial intelligence chatbots ChatGPT and Claude a chance.

While I work for a think tank by day, I am a Roman historian by night and have multiple books under my belt. Recently, I was trying to track down an ancient citation that I was sure existed. I politely asked AI assistant Claude to help find it for me, and the chatbot claimed that it didn’t exist in any literary source. I then found it in the form of a stone inscription, and shared my findings with Claude.

It responded, “So to correct my earlier responses: the evidence […] is epigraphic, not from a literary ancient source — and [your finding] is precisely that evidence.” It continued, “I apologize for not having been aware of it earlier.”

Clearly AI chatbots need more development, but I appreciate the computer-generated humility. Despite this error, they can still serve valuable purposes, and I decided to give Claude another shot. I asked it to track down a different citation. While I didn’t get a great answer, I liked its candid and cautious response.

Claude wrote, “Honestly, I’m not confident enough to name a specific source without risking giving you another incorrect citation — and given the pattern of our conversation, I’d rather admit that than send you chasing a phantom reference.”

I appreciated it making me briefly feel intellectually superior as well as its timidity, which gave me a chuckle. I understand that its humility is not real, but computer-programmed and generated. Nevertheless, if I were to hope for an AI future, I would prefer one with chatbots that exude this type of kindness and humble behavior, but also greater accuracy. Claude and ChatGPT have not reached a satisfactory level yet, but they probably will eventually, judging by other AI applications right here in Georgia—highly autonomous vehicles.

My day job takes me to the Georgia state Capitol regularly, and while in Atlanta, I have witnessed numerous autonomous vehicles—all of them operated by Waymo—navigating the streets safely. If you have never seen one in the wild, then it is a bit of a shock to see cars with empty driver seats scooting by, but don’t let that spectacle alarm you.

Thanks to much more developed AI technology and a host of sensors, these autonomous vehicles have an enviable safety record. “The Waymo Driver exhibited significantly better safety performance, with an 88% reduction in property damage claims and a 92% reduction in bodily injury claims compared to human-driven vehicles,” according to Reinsurance News.

Put simply, on average these autonomous vehicles are involved in far fewer wrecks and far fewer wrecks involving injury or death. And they have logged an amazing number of miles, too. Combined, Waymo vehicles have driven well over 170 million miles.

As more people adopt autonomous vehicles, the safety record could become even better. That’s because they don’t drink and drive, get distracted or fall asleep at the wheel, which will save lives, prevent property damage and alleviate Georgia’s bedeviling traffic. Beyond this, they can help give independence to people with disabilities.

There are myriad other applications for AI, and as time progresses, many others will be uncovered as AI continues to develop. While naysayers disparage AI and predict an apocalyptic future with it, right now it increasingly looks like it will be an indispensable tool. Again, I don’t know the future, but I hope the AI of tomorrow will be at least as humble as Claude and as safe and helpful as Waymo.