AI on the Horizon: Predictions, Challenges, and Citizen Involvement
As AI rapidly evolves, governments are faced with the challenge of staying ahead -- while balancing caution and the need for innovation
Welcome to the latest edition of TL;dr from Good Government Files, where we kick off September with our regular focus on Artificial Intelligence. From Eric Schmidt’s bold predictions about the future of AI to the U.S. government’s push for citizen-driven evaluations of AI systems, and the public sector’s ongoing struggle to grasp generative AI, these articles highlight the urgency for governments to not only understand AI but to actively engage with and shape its development. Whether you’re just getting your feet wet or already navigating the complexities of AI, these insights will help you stay ahead of the curve in this transformative space.
Schmidt's AI prediction: the next two years will shock you
The Exponential View reported on a recent discussion at Stanford, where Eric Schmidt, former Google CEO, provided a candid and somewhat alarming view of the future of AI. He highlighted the unprecedented speed and uncertainty in AI’s development, predicting the next two years could bring dramatic changes that are difficult to predict. Schmidt emphasized the growing scale of AI models and investments, which could have significant geopolitical implications. His insights suggest that even industry veterans are struggling to keep up with the rapid pace of AI advancements. Here are the takeaways.
Unprecedented Uncertainty: Schmidt’s shifting views on AI within just six months illustrate the volatility in the field. As writer Azeem Azhar put it, “It took just six months for a U-turn. That is the degree of uncertainty.”
Unprecedented Speed and Disruption: The convergence of large context windows, AI agents, and text-to-action capabilities will likely accelerate the pace of change. Schmidt noted, “These ‘will have an impact on the world at a scale that no one understands, yet much bigger than the horrific impact we’ve had by social media.’”
Unprecedented Scale of Investment: The financial and energy requirements for advancing AI are immense. “We as a country do not have enough power to do this,” Schmidt said, highlighting the potential geopolitical consequences of the enormous resources needed.
The U.S. Government Wants You—Yes, You—to Hunt Down Generative AI Flaws
The U.S. National Institute of Standards and Technology (NIST), in collaboration with the nonprofit Humane Intelligence, is launching a series of nationwide contests to encourage the public to hunt down flaws in generative AI systems, Wired reports. The initiative builds on the red-teaming exercises held at the 2023 Defcon conference, aiming to increase transparency and accountability in AI technologies. The contests are open to everyone, not just developers, with the goal of democratizing AI evaluation and making it accessible to a broader community. Here are the takeaways.
Broadening Participation in AI Evaluation: The initiative aims to involve a wide range of people in assessing AI, not just experts. “We want to democratize the ability to conduct evaluations and make sure everyone using these models can assess for themselves whether or not the model is meeting their needs,” says Theo Skeadas from Humane Intelligence.
Expanding the Red-Teaming Model: The contest builds on the successful red-teaming exercises at Defcon, where participants were encouraged to find weaknesses in AI systems. “NIST's ARIA is drawing on structured user feedback to understand real-world applications of AI models,” according to Rumman Chowdhury of Humane Intelligence.
Encouraging Transparency and Accountability: The effort includes upcoming “bias bounty challenges,” where individuals can be rewarded for identifying problems in AI models. Skeadas emphasizes the importance of inclusivity, stating, “Policymakers, journalists, civil society, and nontechnical people should all be involved in the process of testing and evaluating these systems.”
What is generative AI? Most public sector workers don't know
A recent survey highlighted a significant knowledge gap in the public sector regarding generative AI, with only about a third of the workforce understanding the technology, reports Route 50. Despite its growing use in various government functions, adoption remains slow due to confusion and a lack of clear policies. Experts suggest experimentation and proper training could help bridge this gap and lead to more effective use of generative AI in government. Here are three takeaways.
Knowledge Gap in Generative AI: The survey found that only 37% of the public sector workforce understands generative AI well. There is still a lot of “confusion” around generative AI and its capabilities among workers, says Jennifer Robinson a global government strategist at SAS who is also mayor pro tem of Cary, North Carolina.
Slow Adoption Due to Overabundance of Caution: There’s perhaps an overabundance of caution in adopting generative AI. As Robinson explained, “One thing that governments don’t want to do is they don't want to get into trouble for stepping out ... governments, at least local governments, always want to be first to be second; they want to see somebody else do it. Then they want to know there were no pitfalls [before] they’ll go ahead and jump into it.” Just over half of governmental organizations have a policy on generative AI, which may also contribute to the hesitation in using the technology.
Importance of Training and Experimentation: To overcome these barriers, Robinson emphasized the need for hands-on experience, saying, “Once you get into something, you understand where those vulnerabilities are, and it makes you better able to say, ‘This is what we need to impose upon ourselves.’”
In Other’s Words
Economics professor Tyler Cowen on raising the aspirations of others:
At critical moments in time, you can raise the aspirations of other people significantly, especially when they are relatively young, simply by suggesting they do something better or more ambitious than what they might have in mind. It costs you relatively little to do this, but the benefit to them, and to the broader world, may be enormous. This is in fact one of the most valuable things you can do with your time and with your life.
Author James Clear on effective communication:
Communication is about what is received, not what is intended. If there is a gap between what you are saying and what they are hearing, you have to find a new way to say it.
Washington Post columnist and conservative author George Will on on the changing nature of the American presidency:
In 1933, President Franklin Delano Roosevelt “gave a fireside chat in which ... he began by saying, ‘My friends.’ Part of his genius was he knew that he was talking to people in their living rooms, because by then, most people had a radio. But it’s inconceivable that George Washington would have said ‘my friends,’ because they weren’t his friends. And more to the point, George Washington’s auditors didn’t want him to be, didn’t expect him to be their friend. They wanted him to take care that the laws are faithfully executed. (That’s) not good enough for the modern president. He wants to be our friend. And this emotional intimacy that has now been established with the chief executive of the United States is infantilizing and unhealthy.”
Onward and Upward.