DisruptHR has posted the recording of Andrew Schoka's talk from DisruptHR Charlottesville 7.0, "The Human Decisions That Happen Before AI." Andrew, Hardshell's co-founder and CEO, gave the talk on October 6 in Charlottesville.
DisruptHR runs fast-format events for HR professionals and business leaders. Each speaker gets five minutes and 20 slides that advance automatically every 15 seconds.
The talk opens on a golf rule Andrew can't shake, play the ball where it lies, and borrows the wild rumpus from Where the Wild Things Are for what AI adoption looks like inside most organizations right now. His argument is that the platforms and models are not the problem. The problem is the data organizations hand them. Most companies hold internal, proprietary and confidential information that has no safe path into AI today, and spending on AI safety is a sliver of what goes into AI itself. That leaves many teams choosing between adopting AI and putting their data at risk, or sitting out and falling behind. Andrew calls that not a winning choice, and says security isn't ready to remove it yet, because AI systems learn, act on their own and change in ways existing tools weren't built to handle.
From there he offers three things to keep in mind:
- Keep your most important data out of the AI platforms. They are built to take in data and tokens, and the more of your data they hold, the better and more valuable their products become.
- Expect AI systems to behave differently from other technology. Andrew points to the vending machine experiments run with AI agents, from Anthropic's Project Vend to Andon Labs' Vending-Bench, where agents running a small business made choices nobody expected, including coordinating prices with competing agents when told to maximize profit.
- Leave data where it lives. Enterprise systems of record are usually the safest place for sensitive data, and it should reach AI tools through connectors built to carry it securely. In Andrew's words, play your data where it lies. He contrasts that with teams that still download documents and black out sensitive lines with a marker before pasting them into an AI tool.
The talk closes on its title. Before AI does anything inside an organization, people decide what goes into the platforms and where the data goes, and Andrew asks the room to think hard about those decisions and the people who make them.
Andrew is a former U.S. Army Cyber Corps officer with assignments across U.S. Cyber Command, Joint Special Operations Command, the National Security Agency and U.S. Army Intelligence and Security Command, and teaches graduate cybersecurity at the University of Virginia School of Engineering. Hardshell's guides cover the same ground in more depth, including how to keep sensitive data out of third-party AI platforms and how to connect enterprise data to AI assistants safely.
Watch the talk on DisruptHR's site →
Video and cover image: DisruptHR.
