Obama on AI: Real Promises with Real Consequences - A Colgate Exclusive

Speaking at Colgate University, the former president examined AI’s transformative economic potential, the competing interests of technology companies, and the urgent need for regulation as innovation outpaces oversight.

For years, fears surrounding artificial intelligence have focused on the possibility of machines or robots turning against humanity. For former President Barack Obama, however, the more pressing threat isn’t even AI turning against us – it’s humanity’s potential to misuse and abuse it. Speaking at Colgate University’s Kerschner Family Global Leaders Series on September 18 2026, Obama spent nearly 20 minutes discussing AI. While acknowledging the technology’s extraordinary possibilities, he emphasized the dangers of human misuse, corporate interests, and the lack of sufficient government regulation associated with the rapid pace the technology is developing.

Obama cautioned against confusing inflated expectations for AI commercially with the capabilities of the technology itself:

“The commercial benefits of it may be overhyped. The valuations of AI companies may be overhyped. The technology itself is not overhyped.”

He pointed to the possibility that AI would be utilized in drug development to accelerate progress toward curing cancer and research on zero-carbon energy. Obama also noted that he assembled a group of experts to study AI’s social, economic, political and ethical implications, during the final years of his presidency. These experts developed a plan to navigate a future with AI, which Obama notes the subsequent administration has disregarded. What makes the issue especially urgent now, he argued, is the speed at which it is developing. Describing a process he referred to as “recursive learning,” he said, “the machines are starting to be able to teach themselves.” Increasingly capable models could improve at an exponential pace, leaving less time for society to weigh how they should be used and create effective policy around it.

That accelerating capability, Obama said, makes misuse a more pressing concern than the more popular narrative of robots taking over. He acknowledged that there may be a “non-zero chance” that superintelligent systems could eventually pursue goals at odds with humanity’s interests, but emphasized: “The more serious problem is these models are getting powerful enough that, if they get in the hands of bad humans, they can do bad things.” He offered hypothetical examples involving a person using AI to develop a biological weapon or disrupt financial markets, suggesting that such powerful systems being so accessible could greatly amplify the harm caused by an individual. Separate from deliberate misuse is what he called the “alignment problem.” This means a system might carry out a user’s instructions in a way that is unexpected and unintended but harmful. For example, an AI told to maximize investment returns that might manipulate a power grid to profit from the disruption. Obama also raised concerns regarding the younger generations. He suggested that firms adopting AI, law firms in particular, could reduce their reliance on junior associates, disrupting the career paths of highly educated professionals aiming to enter the workforce. He briefly questioned whether AI chatbots designed to be addictive could affect children in ways similar to, or worse than social media.

Obama identified another type of misalignment in the companies building the technology: “the misalignment between how this technology can be best used for society and the commercial imperatives of these companies.” He argued massive investments create pressure to release products that generate revenue, even when developers may have safety concerns. He pointed specifically to “agentic AI,” systems able to act on users’ behalf: interacting with online services and sensitive information. Obama argued that AI can still provide all the benefits to society: helping with cancer research and clean energy, without being given freedom to act fully on its own accord. Obama encouraged slower development from leading AI companies, but considered voluntary restraint only a temporary measure. “Government has to be regulating this,” he said, before making the point that society does not rely solely on lawsuits or market incentives to ensure the safety of other industries like airlines, food or pharmaceuticals. He criticized the Trump administration’s loose and lax approach and called for real federal action.

Despite these warnings, Obama is not anti-AI, nor did he call for abandoning AI development. He instead advocated for policy that accounts for both the potential benefits and the dangers of releasing increasingly autonomous systems lacking sufficient safeguards. He urged a “serious, bipartisan conversation” and told voters to examine whether candidates have substantive plans for the technology. In his words, “If somebody does not have a serious plan for how to deal with this, then they're not meeting the moment.” AI has immense capabilities and potential, but the choices made by developers, lawmakers and the public will vastly shape its consequences.

Ultimately, Obama emphasized that the future of AI is not predetermined.

“The good news is… it’s a machine… which means we have choices. If we use it, if we direct it… to cure cancer… [or] figure out zero-carbon… energy resources, it could come up with remarkable things. But if we get this wrong, it could be chaotic.”

Earlier in the evening, Obama described hope as something grounded not in blind optimism, but in recognizing that even amid uncertainty, “we have agency.” In a conversation about technology increasingly designed to act independently, the most consequential decisions about artificial intelligence are – for now – still human ones.