The Biden administration recently issued an extensive Executive Order outlining plans to lead the way in ethical artificial intelligence. However, framing this strategy as a grand vision may be overly optimistic, as its practical impact is likely to be incremental at best.
There is no doubt that the proposals are ambitious. The administration aims to implement rigorous safeguards to prevent potential AI pitfalls, including safety risks, discriminatory biases, job disruption, and more. Requiring disclosure of safety assessments for high-stakes systems would promote accountability, while authenticating AI content would help combat the spread of disinformation.
However, lofty aspirations alone do not guarantee real-world results. Critics point out that previous federal efforts to address algorithmic unfairness have been toothless. Progress on easing visa restrictions for AI experts faces immigration policy challenges. And without key legislation in the offing, the limits of executive authority may constrain sweeping change.
The administration is also juggling a long list of priorities. From privacy to labor rights to global cooperation, the strategy aims to balance a variety of interests. But trade-offs are inevitable, and trying to tackle everything at once risks diffusion and overload.
Grand visions may grab headlines, but execution is what matters. With congressional support lacking, regulatory pushback likely, and the AI landscape constantly evolving, there are many potential roadblocks to the White House’s aspirations. As a result, while the sweeping order suggests promise, the scope of real progress remains uncertain.
For now, it may be prudent to err on the side of caution before declaring the strategy a watershed moment rather than a preliminary patchwork of incremental policies. Only time will tell whether the administration’s ambition will catalyze real achievement or remain mere lofty daydreams.
[Also read a recent HBR article: 3 Obstacles to Regulating Generative AI by Andrew Burt.]
Cheat Sheet: Executive Order on Safe, Secure, and Trustworthy AI.
I. AI Safety and Security Standards
- – Require sharing of safety tests for high-risk AI systems
– Develop standards and tools for evaluating AI safety
– Protect against biological risks
– Authenticate AI content to avoid deception
– Fund advanced cybersecurity AI tools
Implications: Ensuring safety and security is crucial for any team building AI systems, especially for high-risk applications. Following standards and best practices will build trust.
II. Privacy Protection
- – Support privacy-preserving techniques
– Strengthen privacy research and technologies
– Evaluate federal use of commercial data
– Develop privacy guidelines for agencies
Implications: Protecting privacy is a rising concern with AI systems that utilize personal data. Following guidelines and using privacy-enhancing methods will be increasingly important.
III. Equity and Civil Rights
- – Provide guidance on avoiding discrimination
– Address algorithmic discrimination
– Ensure fairness in criminal justice AI
Implications: Teams must proactively address potential biases and inequities in AI systems to avoid perpetuating discrimination.
IV. Consumer, Patient, and Student Protection
- – Advance responsible healthcare AI
– Shape AI’s potential in education
Implications: AI providers should prioritize benefiting consumers and patients while protecting them from harm.
V. Worker Support
- – Develop AI use principles to aid workers
– Study AI’s labor impacts
Implications: Responsibly managing AI’s impact on work will allow for broadly sharing benefits.
VI. Innovation and Competition
- – Support AI research and commercialization
– Promote a fair AI ecosystem
Implications: Fostering an open and competitive environment will drive innovation in AI.
VII. Global Leadership
- – Collaborate internationally on AI
– Accelerate vital AI standards
Implications: Coordinating globally will allow sharing best practices on AI governance.
VIII. Government Use
- – Issue AI use guidance for agencies
– Streamline AI procurement
– Recruit AI talent
Implications: Responsible and effective government adoption of AI can demonstrate benefits and best practices.
If you enjoyed this post please support our work by encouraging your friends and colleagues to subscribe to our newsletter:
[Image from Infogram.]