By Frank F Islam & Ed Crego, May 14th, 2026 (Image credits: Tom de Boor, JNCGPT53)

Managing AI development is a civilizational challenge.

— Eric Schmidt, former Google CEO

Nobody knows how humans can stay in control.

— Geoffrey Hinton, “Godfather of AI”

As we observed in our preceding blog, the United States is fast approaching the precipice where AGI (Artificial General Intelligence) could supplant human intelligence and make the rules for us, rather than us making the rules for it. The failure to recognize and to take the necessary actions to ensure that “authentic intelligence” prevails over AI and AGI could result in a democratic doomsday scenario.

According to IBM, AGI is a “hypothetical stage in the development of machine learning in which an artificial intelligence (AI) system can match or exceed the cognitive abilities of human beings across any task.” There are three critical elements to this definition: machine learning, cognitive abilities and any task.

By contrast, authentic intelligence is formed through human learning and development, which goes beyond the shaping of “cognitive abilities” to perform “any task” to include the formation of social, emotional, moral, and ethical values.

Those human values are developed not merely by collecting and analyzing data, but through personal experiences and interactions with others. They are formed in the life-long process of human development, which occurs in stages, from infancy through childhood and adolescence to adulthood.

There are two primary challenges to be addressed to avoid the potentially destructive and devastating downside of AGI. They are to: (1) Ensure that the AGI systems align with human values. (2) Put regulations and rules in place to control the development, dissemination, and use of AGI and AI systems.

Malihe Alikhani and Sabit Hassan, in their excellent article for the Brookings Institution, titled “Hype and harm: Why we must ask harder questions about AI and its alignment with human values,” point out:

This shift in AIʼs role, from supportive assistant toward autonomous agents, raises critical questions about readiness of the technology and its alignment with human values. Dr. Alondra Nelson, former director of the White House Office of Science and Technology Policy, introduced the concepts of thin and thick alignment. Thin alignment refers to AI systems superficially meeting human-specified criteria, whereas thick alignment emphasizes a deeper, contextual understanding of human values and intentions.

Through their research, Alikhani and Hassan found that using an active learning approach, employing multimodal agents to retrain AI models to handle uncertain or underrepresented scenarios demonstrated significant improvements in safety and fairness, particularly in scenarios involving physical safety and online harms.

There is no question that, as IBM states, “AI-based learning agents,” and “a multi-agent system with each agent handling the part of the task for which they are best suited,” can definitely improve the performance of an AI model and move it closer to “thick alignment” with human values and intentions. It does not ensure, however, complete conformance to, or compliance with, those human values and intentions.

That is why, Alikhani and Hassan advise:

However, technical advances alone will not resolve all concerns around AIʼs integration into our daily lives. Meaningful progress also demands thoughtful policymaking, careful oversight, and continued public dialogue.

Unfortunately, as we discussed in our earlier blog, the Trump administration does not appear to be committed to either “thoughtful policymaking” or “careful oversight.” That is why continued public dialogue, involvement, and investment is essential.

Fortunately, there has been progress in bringing regulatory constraints to bear on AI. The best examples we’ve seen internationally have come from the European Union, and from California here in the U.S.

In September of 2025, California Governor Gavin Newsom signed Senate Bill 53 (SB-53) — The Transparency in Frontier Artificial Intelligence Act into law. In their Brookings Commentary on the law, Malihe Alikhani and Aidan T. Kane report that

SB 53 — applies to developers of the largest foundation models and requires them to produce standardized safety frameworks, incident reporting, internal governance systems, and whistleblower protections. It is an attempt to convert years of voluntary corporate commitments into public accountability that can be evaluated, monitored, and enforced.

Alikhani and Kane state that SB 53 is cutting edge, applies only to the largest foundation models, and establishes a threshold for coverage that no AI system has met yet. They explain:

This threshold is intentionally designed to target only the systems that sit at the frontier of capability development. In addition to frontier models, the law also applies to large frontier developers, companies with annual revenues over $500 million dollars. In practice, this means approximately five to eight companies currently fall under the lawʼs jurisdiction, including OpenAI, Anthropic, Google DeepMind, Meta, and Microsoft.

The European Artificial Intelligence Act (EAI Act or Act) went into effect in August 2024. The EAI Act “is designed to ensure that AI developed and used in the EU is trustworthy, with safeguards to protect people’s fundamental rights.” It classifies AI systems by risk level, and applies rules to each level:

• Minimal risk systems have no requirement
• Limited risk systems have a requirement for transparency to users
• High risk systems must meet strict requirements in areas such as system design, data collection and conformance to safety rules
• Unacceptable risk systems are banned

The EAI Act is being implemented in phases, with the bulk of the Act to be operational by August 2026.

The EAI Act and California’s SB-53 will provide solid reference points when the federal government is ready to put a national policy, regulations, and guardrails in place to manage AI development and utilization. As noted, this will probably not be during the current president’s administration.

This makes the work being done by groups such as Humanity AI, the Brookings Institution Artificial Intelligence and Emerging Technology Initiative (AIETA), and others critical to ensure that human values prevail, and authentic intelligence is not stifled during this interim period.

Humanity AI is a philanthropic coalition formed by ten foundations in October 2025, which have committed $500 million over the next five years to make investments to “establish a people-driven future where AI delivers for humanity, strengthens communities and enhances human creativity.” Those foundations include Ford Foundation, John D. and Catherine T. MacArthur Foundation, and David and Lucille Packard Foundation. The areas they will invest in are: arts, education, humanities and culture, labor and economy, and security.

In the press release announcing Humanity AI, John Palfrey, president of the MacArthur Foundation stated, “The stakes are too high to defer decisions to a handful of companies and leaders within them. Humanity AI seeks to shift that dynamic by resourcing technologists, researchers, and advocates who are united by a shared vision of ensuring AI is a force for good, putting people and the planet first.”

Brookings’ website for AIETA states “The future of AI is not inevitable. Artificial intelligence is transforming how we work, learn, govern, and organize economies. But its rapid evolution often outpaces our ability to understand its full impact….”

And goes on to state, “… Brookings is guiding this transformation by engaging leaders across public, corporate and civic sectors. Through rigorous research and collaborative problem-solving, we deliver evidence-based policy recommendations to help society harness AI’s benefits while minimizing its risks.”

The research areas of focus for AIETA: AI governance and its Impact on democracy, Economic transformation from AI, Opportunities and benefits of AI, AI’s impact on International and Geopolitical Dynamics, and AI Innovations and Future Capabilities.

In conclusion, the pivotal importance of enabling authentic intelligence to rule rather artificial intelligences was highlighted in an AI Summit at Columbia University convened in March 2025. That Summit organized by Columbia’s Data Science Institute featured seven panels of AI faculty members from a range of disciplines.

In an article on the Summit for Columbia Magazine, Paul Hond writes, “The panelists theorized about AI’s potential, limitations, and social impacts and asserted that while AI surpasses the human brain in computational power, it falls short in emotional, social, creative and tactile intelligence.” According to Hond, “Some panelists saw an uphill battle in taming AI.” Even though this battle may be uphill, it must be won for the future of humanity.

Humans with authentic intelligence will be able to respond affirmatively and constructively to the plea “See me, touch me, feel me, heal me.” Machines or robots with Artificial General Intelligence, built only with data, will never to be able to do so effectively or empathetically.

Recognizing this, it is essential for the future of America and Americans that humans control AGI rather than vice versa.