Putting Humanity at the Heart of Synthetic Intelligence

A merger between humans and machines is coming, and it’s not what you might think.

Something remarkable flickered into existence when our ancestors first learned to take knowledge out of their heads and embed it in artifacts. Now, millions of years later, the descendants of those people and those tools are merging into a synthesis of human and machine intelligence. Artificial intelligence and automation act as a platform to support this new synthetic intelligence, and humans give it subjective consciousness. While these deep ties to humanity expose synthetic intelligence to our frailties, they are also our best shot at aligning it with the long-term interests of life on this planet.

Combining Parts into Wholes

Life on Earth is, to some degree, the story of individual parts coming together into new wholes. Economist Brian Arthur calls it “combinatorial evolution.” Microbiologist Lynn Margulis talked about different species coming together in symbiosis and used that insight to explain the ancient origins of plants and animals. The essence of these ideas is that, in the right circumstances, parts can come together to form novel, and more complex, organizational structures. Individual cells come together as tissues, wolves as wolf packs, and engines, engineers, and pilots as airlines. Think of it as parts poured into a new container that then functions at a whole new level of complexity.

Communication and coordination are the crucial ingredients for new wholes to arise from what were once disconnected parts. Intercellular communication allows cells to coordinate themselves as tissues, organs, and organisms. Wolves communicate with howls and body language to coordinate the pack. Engineers communicate through design and pilots through flight instruments to coordinate with the jet engines of an airliner. Now humans are becoming parts in a new whole that is cohering into a synthesis of human and machine intelligence. The containers for this intelligence are a particular type of human organization called “platforms.”

Platforms as the Interface Between Humans and Machines

Though the term “platform” is moving from Silicon Valley lingo into modern vernacular, it is worth clarifying. Platforms are a mix of business model and technology that use digital interfaces to open themselves to large scale external contributions of work and knowledge. Platforms exist in all manner of industries but are most common in the service economy, where Google, Facebook, and Amazon are the most famous examples.

Like all combinatorial evolutions before it, the platform emerged from a breakthrough in communications technology. In this case, it was the Internet, coupled with software interfaces for coordinating work with external stakeholders. As platform operators made their interfaces increasingly intuitive, it became easier for us to serve ourselves. Without talking to a single employee, we can now use them to search vast repositories of knowledge around the world, watch most any show, or book a flight anywhere we want.

These systems are now transforming into humanity’s primary points of contact with machines. No technology touches more people than the platforms of companies like Google, Amazon, Facebook, and Alibaba. It’s not just a question of breadth, either. Our relationship with these systems is deepening as well. Innovations such as Natural Language Processing and emotion recognition technologies make it far easier to communicate with them. This connection will only deepen with augmented reality systems that overlay virtual information onto physical reality and brain-computer interfaces that communicate directly with our brains.

Platforms as Spirals of Applied Synthetic Intelligence

Platforms engage end users on a mind-boggling scale that would be impossible without the automation that their digital interfaces enable. Even more important, though, is the way they use machine learning to make sense of the massive quantities of data from their automated transactions. Machine learning distills those countless individual interactions into abstract mathematical representations of our interactions with the platform.

There’s a strong synergy between our engagement with platforms and the machine learning created by the resulting data. Automation cranks up the scale of engagement, which increases the flow of data, which then fuels more machine learning, and makes the automation smarter and more powerful over time. This positive feedback loop creates an upward spiral in artificial intelligence, but it is narrow in scope: Expedia gets smarter in travel, Spotify in music, and Netflix in shows. Platforms thus represent an extremely effective approach to solving domain-specific problems that relies on an ongoing flow of data from stakeholders.

Machine Intelligence + Human Intelligence

Platforms are more than just businesses though. To see their true revolutionary nature, we need to see them within the context of combinatorial evolution and the possibility that humans are becoming parts in some new whole. We’ve long used corporations and other forms of organization to coordinate learning and work. But the massive scale of platforms and our deepening connection to them represent something qualitatively different: a new synthesis of machine intelligence and human intelligence.

What I mean by the “machine intelligence” of platforms is the intelligence embedded in their automation. It’s no longer just software developers coding systems and letting them run. These systems are increasingly learning on their own by interacting with us. The nature of this machine intelligence is ruthless efficiency, enormous scalability, and unerring precision. Machines are also tireless, and they bring all of these qualities into their “partnership” with us.

The human intelligence embedded in the platforms takes many forms but boils down to what we refer to as subjective awareness, experience, or consciousness. Behind this statement is a set of ideas that we need to grasp if we are to understand our future with machines.

Embedding Human Intelligence in Machines

We could view our relationship with technology in terms of a parent teaching a child, for we are constantly teaching technology what we know. The teaching process takes many forms, but all entail some form of converting tacit human knowledge into explicit knowledge so that it can be embedded into technology. “Tacit knowledge,” is that which subjective human experience can know. “Explicit knowledge” is the subset of what we know that we can actually articulate. So, when we teach technology, we are really explaining what we know in such a way that it can be held in an artifact outside of a human mind.

An arrowhead embeds thousands of years of human experience into a seemingly simple design. A book or a website allows us to embed an infinite range of human experiences in an artifact. Computer software allows us to embed “instructions” into devices so that we can execute those processes automatically at some later point.

We are constantly expanding the range of tacit human knowledge—and thus the scope of human experience—that we can embed in technology. We’ve gone from storing facts in clay tablets, books, and databases to using software instructions to imbue a kind of deferred human volition into our devices. Even intelligence once tightly bound in our biology, such as the ability to recognize faces and spoken words, now rests within machine learning systems. We are, in short, handing over more and more of what we know to machines.

We Don’t Know the Limits to Machine Intelligence

The question this raises is, where does it stop? And here, what we are really asking is whether machines will one day become conscious. The answer, as unsatisfying as it may be, is that we simply don’t know. Given that reality, I believe our work is to set ourselves on a course that best optimizes our chances, regardless of whether machines do some day “wake up.”

There is so much unknown as we search for this path, but I believe that renowned knowledge expert Michael Polanyi offers us a clue when, in describing tacit knowledge, he said, “We know more than we can tell.” Though I doubt he intended it this way, that succinct statement holds great wisdom about our future with machines. What it suggests is that human intelligence is forever pushing the edge of what we know, just as technology constantly works to convert that experience into machine intelligence. This, I believe, is the essence of synthetic intelligence—and a hint for our continued relevance to machines, even if they attain consciousness.

Humanity as the Subjective Core of Synthetic Intelligence

Synthetic intelligence will challenge us to be more discerning in how we focus the human experience. It will help us to concentrate more fully on the cutting edge of human development, while rendering unto technology the things that are best left to technology.

Experience is the essence of a human life. It is what matters most. Were you to wake up one night with your house on fire, your first thought would not be the sterling silver, but your grandfather’s watch and those irreplaceable keepsakes from your childhood. Our most meaningful experiences come in all shapes and sizes: the birth of a child, the thrill of learning something new, the ease of hanging out with a friend by a lake, and the wonder of hearing that song just when you need it the most.

When we dive into these deeper pools of meaning, we discover an emotional tenor, a feeling that can only be described as experiential. It can’t be packaged or converted into explicit knowledge. It is, in short, “more than we can tell.”

There are clearly advantages in having platforms convert our collective experiences into fuel for machine learning. But human experience is more than a commodity to be harvested and abstracted into machine intelligence. We need our systems to understand that while subjective human experience has commercial value, it also has a kind of value that is precious, unique, and intrinsic—not just a means to other ends but an end in and of itself.

With synthetic intelligence we will augment, and increasingly offload, much of our thinking and work to machines. This will leave us free to concentrate on ensuring that the subjective conscious experience that rests within intelligent automation is the very best that it can be. In this way, we shift humanity’s focus to the great work of expanding consciousness. We will sharpen our intellect with the help of machines but opening the warm intelligence of the heart is work that is ours to bear. Our destiny, the destiny of humanity, is to be the heart of synthetic intelligence.

Humanity in a World of Artificial Consciousness

A word needs to be said in closing about a future where machines learn the ultimate trick of experiencing for themselves. We have absolutely no idea what this might mean. We don’t know what machine consciousness might look like or how it would respond to sharing the planet with humans.

Our best chances for peaceful coexistence will be machine consciousness with a built-in appreciation for other forms of consciousness, like ours. Humans have admittedly set a lousy example in the way we’ve treated most other forms of consciousness on the planet, but perhaps there is a way to teach machines to rise above the prejudices of their progenitors.

Life exists in biological ecosystems, but machines exist in economic and political ecosystems. What runs the ecosystems of machines today is the quest for wealth and power, which is what in turn drives the platforms in transforming human experience into machine intelligence. To help ensure that we can share a future with conscious machines, we need to focus on two fronts. First, we must now extend the way we treat human experience so that our platforms value it both commercially as a means toward collective intelligence as well as intrinsically as a kind of inalienable and sacred right. Second, we must now focus on developing humanity so that it is worthy of the tremendous responsibility it is about to experience as the heart of synthetic intelligence.

[Editorial note: on March 10th, I revised this into what is essentially a new article.]

Scroll to Top