rules based system in artificial intelligence :: Article CreatorTruly Intelligent AI Could Play By The Rules, No Matter How Strange
Truly Intelligent AI Could Play by the Rules, No Matter How Strange
To build safe but powerful AI models, start by testing their ability to play games on the fly
By Vinay K. Chaudhri edited by Dan Vergano
A proposed game-playing challenge would evaluate AIs on how well they can adapt to and follow new rules.
Tic-tac-toe is about as simple as games get—but as Scientific American's legendary contributor Martin Gardner pointed out almost 70 years ago, it has complex variations and strategic aspects. They range from "reverse" games—where the first player to make three in a row loses—to three-dimensional versions played on cubes and beyond. Gardner's games, even if they boggle a typical human mind, might point us to a way to make artificial intelligence more humanlike.
That's because games in their endless variety—with rules that must be imagined, understood and followed—are part of what makes us human. Navigating rules is also a key challenge for AI models as they start to approximate human thought. And as things stand, it's a challenge where most of these models fall short.
That's a big deal because if there's a path to artificial general intelligence, the ultimate goal of machine-learning and AI research, it can only come through building AIs that are capable of interpreting, adapting to and rigidly following the rules we set for them.
If you're enjoying this article, consider supporting our award-winning journalism by subscribing. By purchasing a subscription you are helping to ensure the future of impactful stories about the discoveries and ideas shaping our world today.
To drive the development of such AI, we must develop a new test—let's call it the Gardner test—in which an AI is surprised with the rules of a game and is then expected to play by those rules without human intervention. One simple way to achieve the surprise is to disclose the rules only when the game begins.
The Gardner test, with apologies to the Turing test, is inspired by and builds on the pioneering work in AI on general game playing (GGP), a field largely shaped by Stanford University professor Michael Genesereth. In GGP competitions, AIs running on standard laptops face off against other AIs in games whose rules—written in a formal mathematical language—are revealed only at the start. The test proposed here advances a new frontier: accepting game rules expressed in a natural language such as English. Once a distant goal, this is now within reach of modern AIs because of the recent breakthroughs in large language models (LLMs) such as those that power ChatGPT and that fall within the families of Claude and Llama.
The proposed challenge should include a battery of tests that could be initially focused on games that have been staples of GGP competitions such as Connect Four, Hex and Pentago. It should also leverage an impressive array of games that Gardner wrote about. Test design could benefit from the involvement of the vibrant international GGP research community, developers of frontier AI models and, of course, diehard Martin Gardner fans.
But to pass the new test, it isn't enough to create an AI system that's good at playing one specific predetermined game or even many. Instead, an AI must be designed to master any strategy game on the fly. Strategy games require humanlike ability to think across and beyond multiple steps, deal with unpredictable responses, adapt to changing objectives and still conform to a strict rule set.
That's a big leap from today's top game-playing AI models, which rely on knowing the rules in advance to train their algorithms. Consider, for instance, AlphaZero, the revolutionary AI model that's capable of playing three games—chess, Go and shogi (Japanese chess)—at a superhuman level. AlphaZero learns through a technique known as "self-play"—it repeatedly plays against a copy of itself, and from that experience, it gets better over time. Self-play, however, requires the rules of each game to be set before training. AlphaZero's ability to master complex games is undoubtedly impressive, but it's a brittle system: if you present AlphaZero with a game different than the ones it's learned, it will be completely flummoxed. In contrast, an AI model performing well on the proposed new test would be capable of adapting to new rules, even in the absence of data; it would play any game and follow any novel rule set with power and precision.
That last point—precision—is an important one. You can prompt many generative AI systems to execute variants on simple games, and they'll play along: ChatGPT can play a 4×4 or 5×5 variant of tic-tac-toe, for instance. But an LLM prompt is best thought of as a suggestion rather than a concrete set of rules—that's why we often have to coax, wheedle and prompt tune LLMs into doing exactly what we want. A general intelligence that would pass the Gardner test, by contrast, would by definition be able to follow the rules perfectly: not following a rule exactly would mean failing the test.
Specialized tools that operate without truly understanding the rules tend to color outside the lines, reproducing past errors from training data rather than adhering to the rules we set. It's easy to imagine real-world scenarios in which such errors could be catastrophic: in a national security context, for instance, AI capabilities are needed that can accurately apply rules of engagement dynamically or negotiate subtle but crucial differences in legal and command authorities. In finance, programmable money is emerging as a new form of currency that can obey rules of ownership and transferability—and misapplying these rules could lead to financial disaster.
Ironically, building AI systems that can follow rules rigorously would ultimately make it possible to create machine intelligences that are far more humanlike in their flexibility and ability to adapt to uncertain and novel situations. When we think of human game players, we tend to think of specialists: Magnus Carlsen is a great chess player but might not be so hot at Texas Hold'Em. The point, though, is that humans are capable of generalizing; if Carlsen ever gave up chess, he could be a decent contender for the Pentamind World Championship, which celebrates the best all-round games player.
Game playing with a novel set of rules is crucial to the next evolution of AI because it will potentially let us create AIs that will be capable of anything—but that will also meticulously and reliably follow the rules we set for them. If we want powerful but safe AI, testing its ability in playing games on the fly might be the best path forward.
What's Inside The EU AI Act—and What It Means For Your Privacy
In late 2023, the European Union finalized its Artificial Intelligence Act, the world's first comprehensive law governing corporate AI use. The EU AI Act, which takes full effect by August 2026, applies to any company operating in Europe or serving EU consumers, including U.S. Tech giants and startups with overseas customers.
As AI usage becomes more embedded across the public and private sectors, Europe's legislation could pressure American companies to rethink their approach to data privacy, transparency, and human oversight.
Here's what's included in Europe's sweeping regulation, how it might affect U.S.-based business owners, and why it might reshape consumer expectations.
Key Takeaways
The EU AI Act intends to set a global benchmark for responsible artificial intelligence usage by requiring companies, including U.S. Firms, to meet strict standards for transparency, documentation, and human oversight if they serve EU customers. American businesses face real financial and reputational risks if they fail to meet the Act's requirements, especially for high-risk systems like those used in hiring, credit scoring, or law enforcement. Although the U.S. Is not expected to follow suit with a similar federal AI law, consumers will grow to expect AI transparency. Experts say smart businesses should prepare now by aligning with the EU's rules to stay competitive and build trust. What Does the EU AI Act Do? The EU AI Act's main goal is to ensure that companies that develop and use artificial intelligence systems do so safely, ethically, and with respect for consumers' rights and privacy. It classifies AI tools by risk level and applies different compliance rules accordingly. Minimal risk AI systems like AI-powered spam filters and simple video games are largely unregulated. Limited-risk AI systems like chatbots, automated product recommendation systems, and image/video filters and enhancement tools must meet transparency obligations to inform users that they're interacting with artificial intelligence. High-risk AI systems are those used in applications like credit scoring, critical infrastructure, border control management, worker management, law enforcement, and many activities that determine a person's access to resources. These systems face strict documentation, testing, and human oversight requirements, which are expected to go into effect in early August 2026. Unacceptable risk AI systems have been deemed to threaten people's rights, safety, or livelihoods and are banned outright within the EU (with some exceptions). Examples include real-time biometric surveillance for law enforcement or categorization based on sensitive attributes, social scoring systems, and any form of "manipulative AI" that impairs decision-making. This ban has been in effect since February 2025. The Act also includes provisions for "general purpose AI" (GPAI) models like OpenAI's ChatGPT to comply with certain requirements based on their level of risk. All GPAIs must adhere to the EU's Copyright Directive (2019) and provide usage instructions, technical documentation, and a summary of the data used to train their models. Additional compliance criteria apply to GPAI models that "present a systemic risk." While some Big Tech companies have pushed back on the regulation, the European Commission has indicated it's open to amending the Act during a planned review. Why Does the EU AI Act Matter for American Businesses? The EU AI Act applies to any company operating within or serving consumers in the European Union, regardless of where they're headquartered. For American organizations with overseas business partners or customers, the Act could mean significant compliance costs and operational changes for big players and startups. Fines can be as high as 7% of global annual revenue if you use a banned AI application, with slightly lower fines for noncompliance or inaccurate reporting. Yelena Ambartsumian, founder of AMBART LAW, a New York City law firm focused on AI governance and privacy, believes U.S. Companies will start to feel the "regulatory heat" when the provisions dealing with high-risk AI systems go into effect next year. "U.S. Companies must ensure their AI systems meet the transparency and documentation standards set by the EU, which includes providing detailed technical documentation and ensuring proper human oversight," Ambartsumian said. "Failure to comply could result in penalties, market restrictions, and reputational damage." Pete Foley, CEO of ModelOp, an AI governance firm for enterprise clients, added, "U.S. Companies could stand to receive a wake-up call." "They'll all need to reevaluate their AI governance practices and make sure they align with the EU expectations," Foley said. An AI educator, author, and business consultant, Peter Swain, expects the Act's rollout and enforcement to follow the same path as the General Data Protection Regulation (GDPR). "The EU AI Act is GDPR for algorithms: If you trade with Europe, its rules ride along," said Swain. "GDPR already gave us the playbook: early panic, a compliance gold rush, then routine audits. Expect the same curve here." Related Stories Negotiation Secrets That Could Change Your Financial Future The Importance of Strategic Planning Will American Consumers Be Impacted by the EU AI Act? While American consumers might not be directly impacted by the EU AI Act's provisions, experts believe users will get accustomed to higher standards of transparency and privacy by design from EU-originating apps and platforms. Adnan Masood, Ph.D., Chief AI Architect at UST, noted that consumers will gain clearer insight into when algorithms influence decisions, what data is used, and where redress is possible. "Europe is setting baseline expectations for ethical AI, and the resulting uplift in transparency will spill over to American users as companies unify product experiences across regions," Masood said. "Right now, consumers don't know what they don't know," added Swain. "Once Americans taste that transparency, they'll demand it everywhere, forcing U.S. Companies to comply—regulators optional." Will the US Adopt Similar Rules? William O. London, a business attorney and founding partner at Kimura London & White LLP, noted that the U.S. Has taken a more sector-specific and state-driven approach to AI regulation. Still, there is growing bipartisan interest in establishing federal AI governance. While the White House did revise its existing policies on federal AI usage and procurement in April 2025, this is unlikely to lead to a federal regulation resembling the EU AI Act. "Any U.S. Legislation will likely seek to balance innovation with consumer protection, but may be less restrictive to avoid stifling tech development," said London. Ambartsumian noted that AI regulation is becoming more intertwined with politics and industry. "Tech companies have been quite vocal in appealing to the [Trump] administration to exempt them from state laws [on AI]," she said. "The House Energy and Commerce Committee is now evaluating a 10-year moratorium … on state-level laws." At the time of writing, only a handful of states have laws on the books regarding AI usage, including Colorado (which is the most similar to the EU AI Act), California, and Tennessee and several others are considering similar pieces of legislation. While such guidelines can help level the playing field when it comes to AI usage, Foley warns that compliance costs and administrative burdens could strain small businesses' limited resources, especially if they're trying to keep up with nuanced state-specific laws around AI. "It's crucial for policymakers to consider scalable compliance solutions and support mechanisms to ensure that small businesses can navigate the evolving regulatory landscape without disproportionate hardship," Foley added. Regardless of current or pending AI rules in your state, experts say it's wise to start preparing for greater AI transparency if compliance becomes mandatory. "Smart small businesses should calibrate to the strictest standard—the EU—once, then sell anywhere," Swain advised. "Create a one‑page 'Model Safety Data Sheet' for every AI tool—purpose, data sources, and risk controls. It turns red tape into a trust badge." The Bottom Line The EU AI Act is a bold move toward protecting citizens in an AI-driven world. It may very well become a strict model for the rest of the world, or it may get watered down as industries that rely heavily on artificial intelligence fight against regulatory hurdles. Either way, consumers can expect AI-driven services to become more transparent in Europe and eventually, everywhere else.
Comments
Post a Comment