Will AGI Reflect the Best or Worst of Human Nature? 

Google’s recent Responsible AI Progress Report speaks to the question of managing the lightning-fast development of what it calls “the most transformative technology in human history.” Conscious of the prospect of imminent artificial general intelligence (AGI), Google’s AI leadership focuses on three tenets: bold innovation, responsible development and deployment, and collaborative progress together. 

I have always trusted Google and believe its aim is true. I do worry, however, about what AI learns from us as we reveal our true values in the race to adopt and apply its power. Science fiction is full of stories of machines run amok with combative, greedy, or heartless values. It is vital that we avoid this trap. 

What Problem Are We Trying to Solve? 

The BBC published an article last week that saw darkness in Google’s report. The inclusion of language sponsoring work on AI that “supports national security,” plus the omission of reference to prohibiting uses that are “likely to cause harm,” sounds to the Human Rights Watch like Google is okay with AI weapons.  

Does this mean we’re about to program Skynet to terminate humanity? Not at all, but the concern is valid if the problem we’re trying to solve is a geopolitical strategy challenge. AI, whether deployed as countless simple agents or vast reasoning LLMs, will learn how to “win” the war. But are we clear who the enemy is and what victory looks like? 

Similarly, AI deployed to learn what people want to watch or hear is already doing a great job serving up simplicity and entertainment from Netflix to Spotify. And yet, rising adolescent depression, weakening marriages, and misinformation point to the unwanted side effects of machine intelligence tuned to values that optimize for engagement by tweaking our basest emotions. TikTok may be irresistible, but so is heroin. 

Even something so seemingly reasonable as maximizing shareholder value has a dark side. AI focused on learning what assets to trade and how to drive up their prices sounds great if you’re an investor. That said, thinking machines capable of finding every chink in the armor of regulation or gaming partners to their bare minimum break-even terms could easily revive the robber barons of the Gilded Age. 

Humans can be competitive, vain, and greedy. AGI learning what we “want” in purely military, media, or financial domains could end badly for everyone. 

Making the Most of What Nature Provides 

If not the end goals of money, power, and respect, then what?  

How about an indefinitely abundant planet supporting ten billion humans of all shapes and sizes? This is what supply chain people are working toward every day in nearly every industry. Constantly balancing the basics of customer service, quality, and cost is the right kind of value system to teach AI.  

Supply chain functions, including sourcing, manufacturing, and logistics, account for approximately 65% of global carbon emissions, all of which are a proxy for energy, materials, and labor costs. As the organization most accountable for cost-of-goods-sold, supply chain teams naturally embrace carbon reduction, no matter what the political mood of the moment. 

Plus, carbon accounting is only one sustainability vector that supply chain leaders consider. Some, like Ecolab and Schneider Electric, focus on water stewardship; others, like Nestlé and General Mills, invest in regenerative agriculture; and others, including Amazon and Microsoft, are pushing to scale up nuclear energy to power AI without the emissions. 

The Learning Race 

AI is clearly on a tear. Nations are battling to control its future; businesses and investors are jostling to master its potential to drive revenue and profit; media and content creators are racing to master its power to influence. Each has ideals it can’t help but instill in a technology edifice that may one day become sentient. All have the resources to push hard and fast in pursuit of their values. 

Supply chain leaders’ AI mission is simple and even humble. We need to teach AI, in as many instances as possible, to pursue our values by constantly working to balance delivering goodness to customers for the least possible cost.  

Google’s tenets feel right. Be bold, responsible, and collaborative. 

And hurry.