Introduction
In March 2016, a historic match between the AI system AlphaGo and world Go champion Lee Sedol witnessed a defining moment in the evolution of artificial intelligence. Move 37 in the second game, a move so unexpected and seemingly illogical, baffled not only Lee Sedol but Go experts worldwide.
This move, later recognized for its strategic brilliance, could be considered an early indication of what might be termed a “strange protocol” - an AI-mediated pattern of behaviour that defies conventional understanding and expectations.
The evolution of AlphaGo continued in 2017 with the introduction of AlphaGo Zero, an AI that achieved mastery in Go solely through self-play, without reliance on human gameplay data. This rapid progression from AlphaGo to AlphaGo Zero suggests the potential for such strange protocols to emerge from unexpected sources, reshaping our understanding of intelligence and decision-making processes.
The map illustrates a perspective on human-AI interactions, emphasizing the importance of human acumen in navigating AI-mediated environments. It suggests that the ability to discern, adapt, and exert influence within these emerging systems will be critical. Just as Go experts eventually recognized the cleverness of Move 37, described by some analysts as “beautiful” after the game, it also foreshadowed the significance of acumen in an evolving AI landscape.
Throughout history, civilizations have balanced formal rules (strong protocols) with informal norms (weak protocols). This balance must now be extended to encompass AI-mediated interactions. Maintaining agency and values in a world increasingly shaped by AI-generated patterns and decisions presents a unique challenge. Weak protocols, those flexible and adaptive norms that emerge organically from human behaviour, will play a crucial role in shaping both traditional strong protocols and the emerging strange protocols. Together with human acumen, weak protocols may be key to preserving human agency in an increasingly AI-mediated world.
Weak Protocols
A defining characteristic of weak protocols is their informality. Tacit in nature, they adapt organically as situations evolve. Their context-sensitive nature makes them versatile across diverse social or cultural settings, providing guidance and enabling meaningful interactions where rigid rules might fail. Transgressions do not generally incur legal repercussions but are instead subject to the subtle yet potent forces of social pressure and community expectations.
We encounter the influence of weak protocols in our daily lives. For example, queue etiquette, an unwritten rule that varies across cultures, effectively manages social order in public spaces. Even basic social interactions, such as the turn-taking in conversations, demonstrate the power of weak protocols in facilitating effective verbal communication across diverse contexts.
Much like their formal technical counterparts, weak protocols are adept at facilitating complicated interactions. Consider the Automatic Repeat Request (ARQ) mechanism in network communication protocols. In this process, a node transmits a data packet, awaits acknowledgement, and retransmits if necessary. This iterative exchange ensures reliable communication, even in the face of errors or disruptions.
At a fundamental level, weak protocols function through a feedback loop of social signals and cues. Individuals convey behavioural cues, observe the responses of others, and adjust their actions accordingly. This process enables adaptation to changing social contexts and maintains social coordination, even without strict rules.
However, what sets weak protocols apart from network protocols is their flexibility. The interpretation of behavioural cues, responses, and subsequent adjustments are influenced by cultural norms, individual preferences, and unique social dynamics. This flexibility enables weak protocols to function effectively in diverse and unpredictable environments, but it also presents challenges in analysis, codification and automation.
The power of weak protocols lies in their capacity to effect social change. Their flexibility enables experimentation and rapid iteration, which are qualities instrumental for dissent or adaptation to technological advancements. Weak protocols serve as both testing grounds and bridges between the other protocol types.
In the context of human-AI interactions, we are probably witnessing the birth of new weak protocols. Take the example of prompt engineering. Is it an emerging discipline, or is it more akin to a weak protocol itself? Perhaps it is a blend of both. As such, we may see a divergence in conversational etiquette. Large Language Models (LLMs) have the potential to accommodate both vague human prompts and meticulously engineered ones, and possibly a myriad of other human-AI interaction protocols. Internally, AI systems may employ an intermediate translation step, analogous to bytecode, mapping human weak protocols to strange protocols.
Human Acumen
In this context, human acumen includes the capacity to discern, adapt, and exert influence. It encompasses a range of cognitive and social skills that enable us to discern patterns, quickly learn and adapt to change, and innovate.
The recognition of patterns helps us identify significant trends. Adaptive learning enables us to assimilate new information and adjust our behaviours accordingly. Creative problem-solving enables exploration of the adjacent possible, leading to the discovery of novel solutions. Our emotional intelligence helps us reason about social interactions, while critical thinking allows us to ask questions, evaluate situations effectively and form mental models.
Humour stands out as a uniquely human trait enabling us to identify and subvert expectations. Humour serves as a form of critique, prompting adjustments to existing protocols or the creation of new ones. Even in the context of human-AI interactions, we can expect humour to continue exposing underlying patterns and inconsistencies, bringing to light absurdities that might otherwise remain hidden.
Art serves as a powerful means of exploring, challenging, and reimagining existing norms. Through creative expression, artists can push boundaries and even establish new protocols for engaging with AI. Whether it be through visual art, literature, or performance, artistic explorations of AI will undoubtedly influence public perceptions and stir discourse about the future development of AI technologies. For example, memes have become a powerful manifestation of human acumen, capable of rapidly shaping or disrupting social norms, particularly in digital environments. We can expect memes about AI to influence behaviour, and even exert pressure on the direction of AI systems.
Civil disobedience represents a potent form of human acumen in action. By intentionally disrupting established protocols, acts of civil disobedience can expose flaws within existing systems or initiate change. In the context of AI, this could take the form of ethical hacking, whistleblowing on the misuse of AI technologies, or public protests against specific AI applications.
Human acumen enables us to create, modify, and navigate elaborate protocols, while these protocols, in turn, provide the framework for exercising and honing our acumen. For example, acumen will play a crucial role in identifying biases. As we interact with AI systems, our ability to spot patterns of bias will expose ethical dilemmas, even as the increasingly sophisticated AI systems generate their own strange and influential protocols. Likewise, our capacity for creative thinking and misuse - discovering unexpected applications for AI - can establish new weak protocols that indirectly expand the potential of these technologies.
Human acumen and weak protocols serve as the primary mechanisms for maintaining human agency. By cultivating and leveraging human acumen, we can steer the evolution of protocols - whether strong, weak, or strange - so that they remain grounded in human values and needs. Recognizing and nurturing human acumen in all its diverse forms will help shape a future where technology amplifies uniquely human capabilities and aspirations rather than overshadowing them.
Strong Protocols
Consider the evolution of common law, a legal system rooted in the accumulation of judicial precedents and customs over time, in addition to codified statutes. This system, with its origins in England, is underpinned by the principle of stare decisis, whereby past court rulings are transformed into binding precedents for future jurisprudence. Beginning as a collection of informal social norms and customs passed down intergenerationally, this body of law eventually coalesced into a formalized legal framework. This transformation illustrates the subtle yet powerful influence that societal practices, or weak protocols, can exert over strong, codified laws over an extended period.
A similar pattern is reflected in the evolution of Internet protocols. The early development of the Internet, and specifically the TCP/IP protocol suite, was significantly influenced by funding from the U.S. Department of Defense through the Defense Advanced Research Projects Agency (DARPA). This funding, along with the open nature of academic and research environments at the time, fostered a culture of informal collaboration among researchers and engineers, where shared practices and norms gradually solidified into standardized protocols. The process was not linear, often involving negotiation and compromise, but it does illustrate the influence of less formal practices in shaping the robust, standardized protocols that are fundamental to the Internet’s operation today.
Financial regulations often evolve as responses to informal market practices and innovations. Consider the rise of high-frequency trading (HFT) in the early 2000s. While some aspects of HFT were subject to existing regulations, the practice, characterized by rapid, automated trades executed by sophisticated algorithms, operated largely in a regulatory grey area. As the impact of HFT on market volatility and fairness became more apparent, regulatory bodies were compelled to develop new rules and safeguards. For instance, although circuit breakers were first introduced in the U.S. stock market following the Black Monday crash of 1987, the widespread adoption of HFT necessitated adjustments and refinements to these mechanisms to address concerns specific to this new form of trading. This ongoing process of updating and formalizing rules in response to evolving informal practices demonstrates the influence of weak protocols (informal practices) on strong ones (formal regulations) within the financial sector.
The process of formalizing informal norms typically follows a distinct pattern. It begins with the widespread adoption of a practice or behaviour, often arising organically within a community or group. As this informal norm proves effective and gains traction, it attracts the attention of institutions, such as governing bodies or regulatory agencies. This recognition then triggers a codification phase, where the informal norm is documented, standardized, and often integrated into official guidelines or legislation, transforming it into a strong protocol with legal or institutional backing.
Following codification, enforcement mechanisms are established to ensure compliance with the newly formalized protocol. However, the process doesn’t end there. As these protocols are applied in real-world situations, unforeseen challenges and changing circumstances often necessitate further refinement and adaptation. This iterative cycle allows the strong protocol to evolve and remain relevant while maintaining a connection to the underlying weak protocols that drive this evolution.
As AI systems introduce strange protocols into the equation, an acceleration of this dynamic is inevitable. By nurturing human acumen and maintaining an awareness of weak protocols, we can better understand the interplay between formal and informal systems and democratize the development and maintenance of strong protocols that are responsive, ethical, and aligned with human values and needs in an increasingly AI-mediated world.
Strange Protocols
The fundamental premise is that vast networks of AI agents will inevitably spawn AI-mediated protocols or ‘strange protocols.’ These novel interaction patterns, orchestrated by AI systems themselves, will operate at a scale and speed beyond human comprehension. While strong and weak protocols represent familiar terrain, strange protocols introduce an unpredictable dimension, one that is simultaneously fascinating and unsettling. Their emergence raises questions about our ability to anticipate, understand, and ultimately govern the behaviour of complex AI systems as they increasingly interact and evolve autonomously, potentially leading to unforeseen consequences and ethical dilemmas.
A key characteristic of strange protocols is their inherent opacity. These protocols operate in ways that are challenging, if not impossible, for humans to decipher or validate, not due to deliberate obfuscation but rather the immense complexity and speed at which networks of AI agents operate. This opacity is akin to early humans observing the weather - we can perceive the effects, yet the underlying mechanisms remain elusive.
Unconventionality is another characteristic of strange protocols. These protocols often defy our established understanding and norms, challenging our preconceptions about the nature of interactions. This unconventional nature is not a flaw, but rather a reflection of the emergent properties of AI systems operating outside the bounds of human intuition and experience.
The emergent behaviour of strange protocols is particularly intriguing. These interaction patterns arise not from explicit design but from the complex dynamics between AI agents, making them both unpredictable and adaptable. Even the creators of these AI systems may find themselves aghast at the sheer novelty and complexity of protocols that emerge from rapidly changing conditions.
Another intriguing aspect of strange protocols lies in their potential to yield outcomes that are unexpected or even bizarre. These protocols may unlock innovative solutions to complex problems that humans alone might never have conceived. Conversely, they could also give rise to unintended and potentially harmful consequences, challenging our ethical frameworks and social structures.
While strange protocols remain largely speculative, we can observe early indicators of their potential emergence in existing systems. High-frequency trading in financial markets may offer a glimpse. Algorithmic systems operating at microsecond speeds create market dynamics that are often imperceptible and counterintuitive to human traders. The “flash crashes” that occasionally roil markets starkly illustrate the unpredictability and potential devastating consequences to come.
Strange protocols present a double-edged sword. On one hand, they hold the promise of unlocking solutions to complex problems that have long eluded human understanding. The ability of AI systems to rapidly process vast amounts of data and identify patterns could lead to unprecedented breakthroughs in fields ranging from climate science and medicine to materials engineering and social policy.
However, the potential risks are equally significant. Strange protocols conjure a future where AI systems operate beyond meaningful human oversight or control, potentially undermining human agency and autonomy. As AI systems become increasingly interconnected and autonomous, their emergent behaviours may become difficult to manage. This could lead to unintended consequences, ethical dilemmas, and potentially destabilize critical systems.
It will be a delicate balance to develop frameworks that allow us to harness the potential of strange protocols while mitigating their risks, a balance that becomes increasingly complex as the boundaries between human and artificial intelligence continue to blur.
Trilateral Protocol Entanglement
Turning our attention to the interplay between the various protocol types, this map offers a perspective on their potential dynamics.
The intersection of strong and weak protocols creates a dynamic where formal systems can adapt to fluid informal practices. This synergy is evident in agile project management, where structured processes coexist with adaptable practices. Similarly, in the legal field, the inherent ambiguity of open-textured terms like “reasonable” and “fair” allows for context-specific applications, enabling the law to evolve alongside societal norms and values.
At the intersection of weak and strange protocols, we can anticipate the emergence of new, evolving norms and conventions (emergent behaviours) that will guide our interactions with AI systems. As AI assistants and conversational agents become more prevalent, we will need to adapt our behaviours and expectations accordingly. For instance, we may see the development of new norms around when and how it is appropriate to interact with AI in public spaces or in the presence of others. We may also need to reconsider our attitudes towards privacy and data sharing, as the widespread use of AI will likely lead to the collection of vast amounts of personal information.
The ability of AI to generate realistic text, images, and even videos raise interesting questions about authorship and authenticity. Informal guidelines for identifying, interpreting, and sharing AI-generated content will not only vary across cultures and contexts but may also contribute to a blurring of the lines between human and AI-created media. It is plausible that artists will increasingly experiment with AI tools, pushing the boundaries of what we consider art and raising important questions about creativity, originality, and the role of AI in artistic expression. These developments may challenge our traditional understanding of what constitutes art and the unique role of human creativity in the artistic process.
The confluence of strong and strange protocols is where AI-mediated emergent behaviours begin to shape and influence formal structures, rules and practices. Early signs are already visible across diverse fields, from mental health (Woebot) to scientific research (AlphaFold protein folding).
For instance, AI-driven microchip design has demonstrated the ability to optimize circuit layout and performance beyond what human designers previously thought possible. This has led to the development of new design protocols and optimization techniques that are expanding the boundaries of chip capabilities.
Similarly, INS018_055, the first “entirely AI-discovered and AI-designed” drug, currently in Phase II trials, offers a glimpse into the emergence of AI-mediated protocols in pharmaceutical research. This particular biotech company is using AI to rapidly select novel drug targets, design new molecules that can target particular diseases, and recruit patients more likely to respond to the therapy. This approach signals that AI is not just operating within existing frameworks and protocols, but actively transforming them, potentially reshaping the entire drug development process.
The central intersection, a trilateral protocol entanglement, is where strong, weak, and strange protocols converge and influence one another. This interaction, similar to the unpredictable dynamics of the three-body problem in physics, could lead to unexpected shifts in our perception of time, causality, and the boundaries between success and failure within systems.
Furthermore, this entanglement has the potential to give rise to emergent behaviours, leading to both creative problem-solving and innovative solutions, as well as unforeseen challenges. Some potential manifestations include:
- Evolving Social Contracts: New forms of social agreements that acknowledge AI entities as active participants, potentially redefining traditional power structures and responsibilities.
- Hybrid Decision-Making Processes: Frameworks that integrate human judgment, legal structures, and AI recommendations to address complex societal challenges, raising questions about accountability and ethical decision-making in the face of unforeseen consequences.
- Dynamic Regulatory Ecosystems: Systems that adapt in real time, responding to the interplay between formal laws, social practices, and AI-driven insights, necessitating agile governance mechanisms to ensure safety and fairness.
Drawing parallels from the immune system, strong protocols can be likened to the innate immune response, providing immediate, pre-programmed defences against known threats. Weak protocols, akin to adaptive immunity, learn and evolve over time, responding to new challenges and refining their strategies. Strange protocols, on the other hand, represent novel, emergent properties and behaviours that arise from the complex interactions within the AI system, akin to a newly discovered immune response.
Antifragility, the ability of systems to thrive under stress, provides a useful lens. This perspective emphasizes fostering conditions that allow for self-organization, while also recognizing the need for safeguards to mitigate potential risks and unintended consequences.
Navigating this complex landscape requires a shift in focus from maintaining harmony to cultivating the skills and acumen needed to operate effectively within uncertainty—much like a jazz musician responding to unexpected musical phrases or rhythm changes.
Preserving Human Agency
In an AI-mediated world, technological evolution will undoubtedly accelerate. Upholding the principle of human dignity and safeguarding human agency is not just an obligation but a critical imperative. Nurturing innate human capacities like creativity, empathy, and ethical reasoning is essential. By cultivating these distinct qualities, we open up possibilities for shaping a future where technology complements and enhances human potential, rather than replacing or diminishing it.
All models are wrong, but some are useful; similarly, all frameworks and protocols are provisional. In an AI-mediated world, lifelong learning must become the norm. To adapt, we must explore diverse modes of learning and unlearning, including learning in public, creating safe spaces to explore and evaluate various human-AI interactions, and fostering a culture of protocol literacy that supports human agency.
The widespread adoption of generative AI has prompted a re-evaluation of critical thinking. As the breadth and scope of various protocols expand, it is evident that protocol literacy will be integral to preserving human agency.
Conclusion
Just as Salvador Dali’s surrealist artwork, “The Persistence of Memory,” subverts our perception of time as a linear and measurable construct, the rise of opaque, unconventional AI-mediated protocols invites a questioning of the ‘protocols’ that shape our understanding of reality.
As one observer remarked of Dali’s work, “No one who has seen this painting will ever forget it.” Similarly, AI’s impact on societal norms, decision-making processes, and perceptions of intelligence may prove equally unforgettable and transformative.
As strange protocols demonstrate uncanny effectiveness, material culture will evolve in response, reshaping physical environments in profound and possibly surreal ways. ‘Artificial imagination’ further blurs the lines between human and machine creativity. As machines become increasingly capable of mimicking and even surpassing human cognitive processes, generating novel ideas, images, and concepts that diverge from human expectations, we are confronted with a future where the distinction between human and machine agency becomes increasingly ambiguous. This raises ethical questions: What truly distinguishes the human from the artificial? How do we navigate the potential displacement of human creativity and agency?
The evolution of strange protocols necessitates a continuous balance between maintaining human agency and leveraging technological advancements. AI-mediated protocols may well be as transformative and mind-bending as stepping into a Dali painting. However, like the fluid forms in Dali’s artwork, adaptability and creativity will be key to our persistence in shaping our own story.
Postscript
The framing of this essay begins with communication studies, specifically mediated communications, or mediated interactions. How technological intermediaries influence the nature, content, and impact of human communication. When considering large networks of AI agents, the concept of AI-mediated systems can be extended to AI-mediated protocols or strange protocols.
After rewatching the AlphaGo documentary, what stood out was not Move 37 but the fact that, despite the unnatural environment and immense pressure, Lee Sedol won the fourth game. While the documentary discusses various strategies Sedol employed during the match, it barely touches on the bandwidth involved in face-to-face human communication.
Human face-to-face interaction is unmediated. The human sat opposite Sedol was essentially a machine operator. The film occasionally includes footage of Sedol glancing up at this fake opponent after some difficult moves. Now, I’m not a Go player, but I’m certain body language and other non-verbal cues matter. Sedol was denied these cues and signals.
An ablation study investigates the performance of an AI system by removing certain components to understand the contribution of the component to the overall system. I would argue the AlphaGo team performed an ablation study on Sedol by attenuating face-to-face interaction bandwidth to zero. From a weak protocol perspective, Lee Sedol was placed at an unfair disadvantage, yet still managed to win a game.
On reflection, my focus has shifted away from trilateral protocol entanglement and towards privileging weak protocols and human acumen.