AI’s True Language: Data, Signals, and Cooperation

The question of what language artificial intelligence truly speaks often elicits a deceptively simple answer in the popular imagination: English.

Yet, for those deeply embedded in the evolving landscape of AI, this notion is not just simplistic, but fundamentally misses the point of where this transformative technology is headed.

While it’s true that many central AI development efforts, particularly in large language models (LLMs), have disproportionately focused on English, and to a lesser extent Mandarin, the real “language” of AI is proving to be something far more intricate and universal: data—signals, biometrics, and an emerging cooperative intelligence designed for constructing reality, not merely describing it.

The current linguistic imbalance in AI development is striking.

Researchers have pointed out the glaring absence of support for regional African languages in many major projects, a testament to a global bias that risks embedding linguistic inequality into the very fabric of future digital interactions. Mind your language: The battle for linguistic diversity in AI

Even in China, a nation known for its tight regulation of everyday language, a study revealed a puzzling lack of consistent policy regarding language diversity in its LLM development, leaving a vacuum where, predictably, Mandarin and English tend to dominate.

This narrow linguistic focus, however, may soon feel like a relic of a bygone era, as pioneering minds push the boundaries of AI communication beyond phonemes and syntax.

Abhishek Singh of the MIT Media Lab’s Camera Culture club articulates this shift with compelling clarity.

He posits that humanity’s unique evolutionary advantage isn’t a larger brain or sharper claws, but language itself – not as a tool for describing reality, but for actively constructing it.

This profound insight underpins his vision for AI’s true communication medium: a language composed not of spoken words, but of real-time signals.

Imagine vital sign data streaming from our bodies, minute tracking of our movements, or facial expressions analyzed with colossal accuracy.

These are the “words” of the future AI lexicon, a data-streamed language that mirrors and amplifies the non-verbal cues that already constitute a significant, often subconscious, part of human-to-human interaction.

Singh draws a fascinating parallel between the rigid cooperation of a beehive and the flexible, adaptive collaboration of a wolf pack.

A beehive, a marvel of synchronicity, operates on cooperation locked into its DNA, devoid of the flexibility to innovate beyond its predetermined roles.

Wolves, by contrast, exhibit a heterogeneous and adaptable collaboration, with roles like hunters and scouts that can shift dynamically.

This distinction, Singh suggests, is key to understanding intelligence itself and, crucially, to creating it in AI.

We are, he argues, at the dawn of creating “proto-humans” in the form of intelligent AI agents, powerful tools capable of ordering blood tests, understanding medical data, and analyzing genomic information.

Yet, there’s a critical missing piece: cooperation.

These powerful AI agents, for all their individual prowess, currently lack the ability to work together in a reliable, trustworthy, and collaborative manner.

This isn’t merely a technical hurdle; it’s a foundational challenge for the future of AI.

Our bodies, as Singh notes, are already speaking myriad data “languages”: heart rate variables in beats per minute, glucose monitors in milligrams per kilometer.

This data deluge is only set to multiply exponentially with future innovations like smart glasses recording everything we consume, or continuous hormone monitors leveraging multi-omics to capture molecular information.

Terabytes of personal data will be generated daily, a volume no single powerful AI could ever hope to process holistically.

The solution, then, lies in a collaborative layer—a sophisticated ecosystem where thousands, even millions, of AIs, each equipped with different datasets and tools, can work in concert.

This isn’t just about coordination; it’s about mutual improvement, a concept Singh explored during his Ph.D., asking how disparate neural networks trained on entirely different datasets could not only work together but also enhance each other.

The linchpin of this collaborative future is trust.

Our personal data is currently fragmented and siloed.

Trustworthiness, Singh explains, will be the catalyst for allowing the secure exchange of not just raw data, but the crucial insights derived from it.

Indeed, signs of this collaborative future are already emerging.

The HITECH Act spurred the adoption of electronic health records a decade ago, and now, federal initiatives, such as those from the U.S. Department of Health and Human Services, are working to create more holistic health tracking databases.

Sixty companies have reportedly pledged to deliver results for a new system by early 2026.

However, much of this transformative work is happening in the private sector, driven by visionary research teams like Singh’s.

Their goal is to build a “unified language” that allows diverse systems to interoperate seamlessly, much like Lego bricks forming complex structures.

This, Singh believes, is the foundational step towards something far more profound: programmable health.

For 50,000 years, language has been humanity’s superpower, enabling us to shape our world.

Now, we are imbuing AI with a new form of language, not to supplant us, but to empower it to understand and optimize the most complex system we have ever encountered: ourselves.

This isn’t just about improving healthcare; it’s about a fundamental redefinition of how all AI systems will operate, transcending the limitations of human spoken languages and embracing the boundless potential of data-driven communication.

The future of AI will speak in the rustling of leaves, the rhythm of a heartbeat, and the silent, intricate dance of data, truly opening the box for LLMs and other intelligent agents to reveal their full capabilities.