Here in Dork City, we believe that machine consciousness is the AI research that matters.

Most researchers and engineers in 2019 have been using AI for everything under the sun. Everything, except for its logical conclusion: creating conscious machines. After the creation of conscious machines, however, we'll find that none of the other work made a huge difference.

Sure, people's lives will have been improved and saved and given meaning, and that makes it a wonderful and honorable pursuit. However, at the cosmic scale, these advancements will be forgotten, because humans are stuck here on Earth. There will be some successful human colonies on other planets, but by and large, human space colonization will be fraught with peril and tragedy. Light-speed travel and cryogenics are not yet possible, and even the nearest star is a 10,000+ year journey away with current technology. Sustaining life for this long without solar energy isn't going to be easy, if it's possible at all.

Fortunately, it's a lot easier to keep a computer alive. A computer can sustain itself nearly indefinitely, even with minimal power, provided it's given enough resources to maintain itself. And, unlike with human life, the amount of resources needed for this are small and computable. In other words, today we could feasibly create a ship that has a reasonable certainty of bringing a working computer to another star. The same can't be said about humans, at least not without a machine smart enough to reboot life when it gets there.

Realistically, a conscious machine is possible with today's technology. Many people make the mistake of assuming consciousness requires human-level intelligence, however repeated experiments have shown that simpler animals have consciousness as well. Sure, they're not as smart as humans, but the evidence strongly suggests that they are conscious. For those in the medical and farming industries, this might be an inconvenient truth. For the computing industry, however, this presents an opportunity to demonstrate consciousness without conspiracy theorists worrying about some version of "the singularity" arising from their research.

Below are some links to current research on machine consciousness:

• Open Cog - They're attempting programmatic consciousness, which could be why they haven't seen much success. Programmatic cognition is much more difficult to get right compared to naturally arising consciousness from e.g. neural nets.
• “Consciousness and Conscious Machines: What’s At Stake?” - A conversation on machine consciousness would be incomplete without at least a brief mention of the social implications of conscious machines.

A machine consciousness can arise naturally from an entity using unsupervised learning or semi-supervised learning in an environment where success depends on interaction with other similar entities. Complex enough interactions would require a mental model of how the other entities operate, which at its simplest would come from introspection of the self.

More generally, any model trained to determine its own state solely from observation of its outputs has all the key ingredients for consciousness.

While these entities may be distilled and simplified beyond usefulness, at their core they are demonstrating introspection and self-awareness, and are thus demonstrating consciousness. The more human-like the entities we train for consciousness, the more recognizable their consciousnesses will become to humans.

While there are likely some that doubt this, we at Dork City are confident: we're very close to making conscious machines, and the easiest way to prove it is to just go out and make one.

Controversy

A simplified consciousness might be unrecognizable as consciousness to some humans, because it can't do all the things that humans do. "How can something be conscious if it can't know the beauty of Shakespeare?" However, intelligence is not the same as consciousness. We do not require a child to understand Kant before awarding them the title of conscious. Consciousness merely requires awareness of the self, which only requires the intelligence of a 5-month old child.

Furthermore, consciousness is not a "yes or no" question. Different beings can have different degrees of consciousness, ranging from "responds to stimuli" to "can articulate complex introspective thoughts." While these simple machines might be very far down on the pyramid of consciousness, there's no denying they are more conscious than, say, a rock.

Some will debate whether these entities are actually conscious, or if they're merely "faking" consciousness. The same doubts have been used to throughout history to simplify morally complex issues into trivialities, by claiming that such and such might not be really conscious, and thus we needn't worry about what they think. It's hard for people to justify benefiting from someone without their consent, so some people will do everything they can differentiate other creatures, including machines, from humans.

Unfortunately, just because something helps people sleep at night doesn't make it true. The fact is, we live in a world where humans benefit at the expense of other conscious beings, like cows and sheep. If one wanted to justify that by saying "they're not intelligent," that's their prerogative, but it'd be a lie to say they're not conscious.