Month: September 2022

Plot:

Almost Human is a “buddy cop” TV series with a twist: It’s set in 2048, and one of the partners in an android. It is set in an unnamed American city where futuristic technologies deliver both great promise and peril for its citizens–some have lives of luxury, others are impoverished and have been left behind, and criminals have been empowered by the new tools at their disposal.

Detective John Kennex (played by Karl Urban) is a classic, hardboiled cop. He’s hotheaded, traumatized by violent experiences in his past, and struggles to form social bonds with others. Due to a change in police procedures, he’s paired with Dorian (played by Michael Ealy), an android with human emotions and a more balanced personality than Kennex.

The series follows their unlikely partnership and the evolution of their bond, as well as of their unique personal stories, as they investigate crimes together. Every episode pits them against a new criminal or group of criminals who use a different kind of advanced technology.

I thought Almost Human was respectably thought-out and entertaining. Kennex and Dorian had an interesting and often funny personal chemistry, and the other recurring police characters were well-acted. The fictional universe in which it was set showed a high attention to detail in fleshing out the advanced technologies that would be available, as well as their social effects, though as my analysis will show, it wasn’t perfect.

I think the show failed to adequately explore how being an android and living among humans shaped Dorian’s inner world, which would have posed questions of greater intellectual substance to the viewer. At times, he seemed too much like a funny human who could do advanced calculations in his head. The plots also got more convoluted and, frankly, worse as the series went on, probably because the writers were running out of material. Almost Human was cancelled after only 13 episodes. While the show wasn’t spectacular, it would have been nice to see the additional character development and exploration of future technologies that would have happened had it been allowed a full season of 22 – 26 episodes.

Analysis:

Episode 1

Fully convincing androids will exist. During scenes set in the police station and in field missions, androids are almost always present. Aside from their mechanical way of talking and emotionless faces, they are indistinguishable from humans. Dorian is the only android at that precinct who has emotions and a warm personality. While androids will be very impressive by 2048, they won’t be able to mimic humans as exactly as they could in the show.

In my big list of future predictions, I wrote that this would be the case by the end of the 2030s: Combining all the best AI and robotics technologies, it will be possible to create general-purpose androids that could function better in the real world (e.g. – perform in the workplace, learn new things, interact with humans, navigate public spaces, manage personal affairs) than the bottom 10% of humans (e.g. – elderly people, the disabled, criminals, the mentally ill, people with poor language abilities or low IQs), and in some narrow domains, the androids will be superhuman (e.g. – physical strength, memory, math abilities). Note that businesses will still find it better to employ task-specific, non-human-looking robots instead of general purpose androids.

To elaborate, I predict that those kinds of androids will be very few in number by the end of 2039, and will be technology demonstrators and prototypes that get a lot of media coverage at carefully controlled tech company demo events. They won’t be available for any person to purchase, won’t roam around public spaces, and won’t have important jobs like working as police officers.

By 2048, the androids will be better, and aspects of their physiques, intelligence, and capabilities will overlap even more with humans, but they still won’t be able to pass as one of us in normal situations. Their body movements will be clumsier and more limited than the average human’s, probably leaving them with the same overall reflexes, nimbleness, balance, and speed as an elderly human. They will also lack the battery life to function for a whole work day in a physically demanding occupation like street cop. Also, if you could examine one at very close distance, you would see that its skin and other external features were less detailed than those of real humans.

A plausible role for an android in a police station of 2048 would be working at the reception desk. It would be tasked with talking to members of the public who came in, could answer most of their questions correctly, and could summon a human officer with the relevant expertise to deal with questions and issues it couldn’t handle alone. The android would be able to walk around the police station and to physically interact with most things it encountered (e.g. – operate door handle), but it would not be as fast or as coordinated as the average human. It would not have a gun and wouldn’t know how to fight criminals. It’s purpose would be to free up a human police officer for duties more crucial for public safety.

Androids and many other machines will be able to pass the Turing Test and to carry on long conversations with humans and to recognize human emotions and to simulate their own. Their personalities will probably rank somewhere between Dorian’s and the “stiffer” androids assigned to the police precinct.

Shooting an android in the head will kill it. There a scene where a police android is shot in the head and instantly dies. This is unrealistic because it will make the most sense to put androids’ CPUs in their torsos instead of in their heads. Doing such would improve their balance by lowering their centers of gravity, and would make them more robust since their “brains” would have more protection around them since a torso is wider than a skull. Their lack of lungs, hearts, and digestive systems will leave them with extra space in their torsos anyway. For more details, read my blog post What would a human-equivalent robot look like?

To look like humans, androids will still need heads, though their CPUs and other critical hardware won’t be in them.

Episode 2

Criminals will use “DNA bombs” to mask forensic evidence. After a pair of professional hitmen murder a man in a hotel room, one of them leaves a small canister behind that explodes after they leave. It is a “DNA bomb,” and it releases a mist composed of innumerable DNA particles, which attach themselves to all the surfaces in the hotel room, masking whatever genuine DNA evidence anyone left behind. Thanks to this, the police detectives are unable to extract useful genetic evidence from the scene.

This is a creative and probably plausible idea. Mass producing random but complete human genomes and packing them into cell-sized particles that could be sprayed out of a can is probably impossible now, but by 2048, the technical challenges might be overcome. Instead of exploding like a grenade, a DNA bomb might work better if it slowly released its load as an aerosol, like a modern “bug bomb.”

There will be sex androids. One of the “people” involved in the aforementioned murder is a female android built for prostitution. By 2048, I’ve predicted androids will be “adequate” in terms of physicality and duplication of the human body and its movements to perform sex acts on real people, though I doubt the experience will be that satisfying. However, if your senses were impaired by alcohol and the darkness of a closed bedroom, it will be good enough.

Machines will be able to monitor your vital statistics at a glance. In one scene, Dorian the android sees that his human partner’s heart rate has increased, indicating he is feeling sexual attraction to a nearby sex robot. Dorian mentions this to tease his partner. Androids and other machines will have this ability by 2048, as well as the ability to detect other vital information from nearby humans, giving them insights into many things the humans are unconsciously revealing, and perhaps trying to hide.

Machines can already “see” human heartbeats: In 2011, a group of MIT students built a device styled after a bathroom mirror that had a built-in camera capable of seeing “the minute changes in skin tone that occur as facial capillaries fill and empty with the beating of a heart.” The mirror contained a display, which showed a numerical readout indicating the heart rate of the person standing in front of it. By 2048, the technology will be even more advanced. By then, expect some machines to have the ability to monitor multiple vitals at once, including voice stress, pupil dilation, blinking rate, and body language, to create real-time, composite profiles of people’s emotional states, honesty, and healthy. They will be the ultimate lie detectors and empaths.

Episode 3

Androids will have more durable bodies than humans. During a gun battle, a bullet ricochets and hits Dorian in the head. While he is damaged, he stays mostly functional and doesn’t lose consciousness. The wound looks bad enough that it probably would have instantly killed a human had the bullet struck them in the same place.

Androids certainly have the potential to be much more durable than humans, and with 2048 levels of technology, we could build androids that had bulletproof skulls and flesh (at least against pistol and lighter rifle bullets). However, I think fears of robots going haywire and attacking humans will wisely dissuade us from doing that, and the androids that do exist will be no faster, stronger, or damage-resistant than average humans.

In the far future, the sky will be the limit for robot design, however.

Episode 4

Human chemists will be needed to make illegal drugs. This episode focuses on a new synthetic drug being sold in the city. The police try to infiltrate the gang that is peddling it by disguising their forensic scientist as a rogue chemist and having him offer them his services. The gang gives him a chance by taking him to their secret lab and letting him synthesize the drug from base ingredients.

By 2048, fully automated labs will exist, and they will be able to make drugs of any kind without human help. The notion that a talented human’s “special touch” is needed to complete the process will be obsolete. That said, the machinery will still be expensive and the lab setups complex, so only pharmaceutical companies, government agencies, and perhaps well-resourced drug cartels will have them. A lower-level drug gang that only spanned one or a handful of cities would still need humans to do the lab work.

However, in the farther future, automation will create major problems by making it easy for ordinary people to synthesize drugs, or to engage in other illegal activities like building machine guns, committing thefts, or even murders. Remotely killing someone might become as simple as verbally telling a quadcopter drone to find the target, shoot him, and then fly to a distant location and self-destruct to erase the evidence.

Robots will be used as shields. In one gun battle between the police and the drug gang, the gang’s android deliberately steps in front of its boss, and uses its bulletproof body to block incoming fire. The injuries don’t appear to affect the android, and it then physically fights with the police. This was creative, and is also a realistic depiction of how androids could be used in combat situations in 2048 (I also saw this in the movie Chappie, when a humanoid robot was placed in the front of a line of police breaking through the front door of a criminal’s house). While we still won’t trust machines to make life-or-death decisions and won’t give them guns, we’ll have no problem using them as bullet shields, distractors, or medics to carry away injured humans.

Episode 5

Machines won’t be able to perfectly imitate human voices. The police find an audio recording of a recent murder. In it, a man utters a few words before shooting the victim. The forensic scientist matches the voice to that of a man who has been in prison the whole time, which seems to exculpate him since he could not have been physically present at the crime scene (it turns out his clone committed the murder). The forensic scientist then says that the man’s voice could not have been faked at scene by a machine since no technology can mimic a person’s voice so accurately.

While this is the case today, I don’t think it will be true by 2048. Given recent progress in machines mimicking human styles of musical composition and artistry, I think it’s certain that they will figure out how to perfectly imitate individual human voices within the next 26 years.

Episode 6

Each android model will consist of many, identical individuals. In this episode, Dorian meets an android of his same model, and they look identical. This will be the case for reasons of economy: It is cheaper for companies to make long runs of identical products than it is to make each on unique. While there will be one-off, bespoke androids in 2048, most of them will be mass-produced products that come off assembly lines.

That said, robotics companies will make efforts to vary the appearances of their androids in the same way that today’s car makers sell the same model in different colors and option/trim packages. Customers will have choices over hair, eye and skin color, and maybe other biometrics (today’s sex doll industry probably offers insights into what physical parameters will be selectable). However, it’s still common for car owners to encounter vehicles identical to their own on the roads, and so it will be for androids in 2048.

Episode 7

Androids will be able to yell really loudly. During a car chase, Dorian communicates with the criminal vehicle by yelling at it with the same volume that a human could only achieve with the help of a bullhorn. We already have tiny, simple devices like smoke alarms that can generate noises louder than human vocal cords can produce, so there should be no technological or financial hurdle to gifting androids in 2048 with the same capabilities. It might be a useful, nonlethal defensive feature that they could use to repel bad humans (perhaps in defense of their human owners) or to summon help in emergency situations.

If we ever get into a war with intelligent machines, they will probably make use of sound warfare during engagements. Loud, startling noises distract and scare humans and make it harder for us to communicate with each other. Machines, on the other hand, would be little affected.

There will be tiny, disposable cameras. In the episode, a perverted criminal paralyzes a victim, locks and explosive collar around his necks, places thumbtack-sized cameras in the victim’s car, and then leaves the scene. When the victim awakens, his panicked, final ordeal is filmed by the cameras and the footage streamed to the internet for people to watch, before the criminal remotely detonates the bomb, killing the man.

With the rate at which electronics are shrinking and dropping in cost, cameras like this will be available by 2048. As in the episode, they will be cheap, single-use devices with adhesive sides, allowing them to be stuck to surfaces, and they will have wireless transmission capabilities and enough battery life to function for a few hours.

Episode 8

There will be guided bullets. In this episode, a team of assassins is using an advanced military rifle that fires guided bullets to kill people in the city. I think guided bullets will be reliable, affordable, and effective by the 2050s, though they won’t be able to perform the sharp turns or to linger in the air like the ones in the show could. One or two degrees of course change per 100 meters of bullet travel is more like it. The shooter would still need a clear line of sight to his target, and would still need to carefully aim the weapon at it. The guided bullets would turn near-misses and off-center hits to nonvital areas into consistent headshots, making average shooters as effective as today’s trained snipers.

That said, small, aerial drones armed with off-the-shelf guns or small explosives could let assassins in 2048 do remote, autonomous killings of people, like those depicted in the episode. By then, a variety of technologies that only big companies and government agencies have now will be more advanced and available to the public. It will be relatively easy to equip a drone with sensors, including cameras loaded with facial recognition algorithms, that allow it to track down specific humans and kill them. In other words, by 2048, assassins will be able to use high-tech weapons to remotely kill people as happened in the episode, but the weapons won’t be guided bullets.

There will be a technology that lets people erase specific memories. A woman who learns that she is the assassins’ next target hatches a plan to make them leave her alone. They want her dead because she knows their identities, so she visits a black market doctor to have him use a machine to delete her memories of them. She plans to videotape the procedure and send it to the assassins as proof.

Our understanding of how the brain stores memories is poor, and while it will surely be better in 2048, I doubt there will be medical procedures that can erase specific memories. Part of the reason is that individual memories are not stored in discrete locations within the brain–any one memory is spread out among neural pathways distributed throughout a brain. Moreover, even if you could somehow erase one memory, the changes it would make to the pathways would probably erase or diminish memories of other things.

Current research into treating PTSD could lead to therapies where people take drugs in controlled clinical settings, while focusing on bad memories, to diminish them. None of the drugs have proven successful yet, but by 2048, it’s plausible at least one could be approved. However, I doubt it will be anywhere near as effective as the memory-erasing machine featured in the episode.

Episode 9

Combat robots will play dead sometimes. Hoping to gain access to the police station’s heavily guarded evidence room, an evil android kills a random woman in public, knowing that the police will quickly arrive. Once they do, the android tries attacking them, provoking their gunfire. The evil android collapses after the first bullet impact and pretends to be dead. The ruse fools the police, who then take the android to the evidence room for later examination to determine why it killed the woman. After a few minutes, the evil android reactivates itself and starts running around the room.

This kind of ingenuity is something we should generally expect from AGIs. “Playing dead” is a specific tactic that will probably become common among combat robots. Unlike humans, machines will be able to totally shut down their life functions for temporary periods, making it impossible for observers to tell if they were actually dead. Feigning death would be a valuable tactic since it would let them do surprise attacks on unsuspecting enemies (i.e. – it jumps up and attacks you from behind right after you walk by it), or to escape after the enemies left the area. Moreover, the fact that robots are capable of playing dead will force enemies to totally destroy hostile combat robots before proceeding, slowing them down and forcing them to expend more munitions.

Episode 10

Advanced human genetic engineering will start in the 2020s. In this episode, it’s revealed that a small but highly visible minority of people are genetically engineered. Several young adult characters, including one of the police detectives, were engineered at conception to have ideal combinations of looks, intelligence, and health. These highly modified people are nicknamed “Chromes.” Based on their ages and the fact that the show is set in 2048, we can conclude that human genetic engineering became routine for rich people in the 2020s. This won’t happen.

The first genetically engineered humans (both female) were created in China in 2018. Instead of being genetic supergirls full of hundreds of DNA tweaks, the twins only had alterations to one gene called “CCR5.” The changes were meant to confer enhanced natural resistance to HIV infection, which was especially useful for them since their father has the virus. Though the geneticist’s intervention did alter their genomes, it’s unclear whether the targeted gene was changed in the desired way. One or both of them might actually have not benefitted from the procedure, or might even be worse off thanks to unwanted alterations to other genes. Only time will tell.

This struggle to change just one gene in a human embryo shows how behind schedule our technology is in creating highly engineered people like the Chromes. Moreover, there’s still a huge social stigma in Western countries about genetically modifying humans.

It’s more realistic that, by 2048, human genetic engineering will start becoming common among rich people. Instead of being able to customize your offspring in every respect and to make them the “total package” of looks, smarts, and athleticism, you might be able to change ten genes, which would only give them slight advantages over naturally born people. It won’t count as “advanced” genetic engineering. In fact, in 2048, IVF embryo selection might actually provide more benefits than genetic engineering.

Professional advice will be available anywhere. While investigating a suspicious death, the police question a man at his home. Concerned about his legal rights, the man summons his lawyer via telepresence to mediate. The lawyer appears as a hologram in the middle of the room, and repeatedly interrupts the conversation between the other parties in ways meant to protect his client.

I doubt 3D holograms like that will exist by 2048, but I’m sure that other forms of telepresence will let lawyers and other people like doctors, therapists, and personal trainers interact with and help us in the real world almost anytime. Additionally, even if true AGIs don’t exist by then, narrow AIs will be advanced enough and good enough at natural language to accurately mimic other humans, and to render useful professional advice as a human with those skills would. This kind of access to professional advice will partly level the playing field between people with different personal resources, and change society in many other ways we can’t imagine now.

That means the police questioning scene will be fundamentally accurate for 2048, though the lawyer would only be visible on a video display in the room, or as a 3D rendering that could only be seen with the aid of augmented reality glasses.

Episode 11

It will be legal for machines to kill people. In this episode, hackers remotely take over a home security system belonging to a rich couple. As a result, an automated machine gun turret shoots the husband to death. It is later revealed that this was retaliation against the family because the same computer-controlled machine gun had killed a harmless teenager who had trespassed on the yard a year earlier.

By 2048, the technology will exist to build a home security system that could tell trespassers apart from residents and then shoot them. However, it will be illegal to possess, and only people like dictators and crime bosses will have them. Humans will strongly resist the idea of giving machines the right or ability to kill other people without human input (this is also why android cops won’t have guns), which is also why armed police, jurors, and judges will be among the last jobs to be automated.

The big exception to this will be in the military sphere. By 2048, at least one major military will be using some type of combat robot (whether it is airborne, seaborne, or terrestrial) that is empowered to fire on human enemies autonomously. While I expect there will be a global ban on autonomous killer drones, it will ultimately be discarded once the technology gets good enough and cheap enough. The potential military advantages will be too great to resist, and enforcement of any ban will be nearly impossible since killer robot factories will be much easier to hide than, say, nuclear weapons facilities.

Episode 12

Nanomachines will change human bodies from the inside. In this episode, a deranged man who hates his own appearance kills people so he can get their DNA samples and then alter his own genes so he gains specific, desirable physical features from them. A black market surgeon helps him with this by performing an experimental procedure in which nanomachines programmed with the victims’ DNA are injected into the criminal’s face. The nanomachines then alter the tissue in the criminal’s face so they match the facial features specified in the victims’ DNA.

First, if you wanted to steal another person’s DNA in 2048 or today, you wouldn’t need to kill them; you would only need to grab a discarded plastic cup they drank out of, or a tissue they blew their nose into, or something like that. People shed their DNA constantly.

Second, in the longer run, we’ll understand what every part of the human genome does, leading to the creation of something like a huge catalog of outward human features (like nose shapes and eye colors) matched with the combinations of genes that produced them. If you wanted a nose job, you could just look at the catalog to find one you liked instead of walking all around a city staring at strangers’ noses until you found a good one. Then you could alter your nose genes accordingly.

Third, there’s virtually no chance that nanomachines will be advanced enough to do plastic surgery on people by 2048. Progress developing nanomachines has happened at a snail’s pace, and the few that do exist have no useful capabilities. In theory, nanomachines will these advanced functions could exist someday. After all, the existence of flesh-eating bacteria and of bacteria that stimulate other cells’ growth show that nanoscale organic machines can alter how much tissue there is in part of an animal. A big and unsolved problem is controlling the behavior of the nanomachines once they’re injected into a person’s body.

By the end of this century, a plausible nanotech-based plastic nose job would involve the patent having his head held tightly in place with restraints while nanomachines (either of fully synthetic construction or highly modified bacteria) were injected into his nose with very fine needles. Some kind of external device, maybe using radio waves, pulses of light, or magnets, would activate the nanomachines, carefully control their activities, and keep them in very specific parts of the nose. One square millimeter at a time, the cartilage and bone in the patient’s nose would be destroyed or built up, slowly changing its overall shape.

Due to safety concerns and probably also to the limitations of the technology, the nanomachines would either be removed or would stop working after a short time and disintegrate. Multiple sessions involving the technique, spread out over weeks so the plastic surgeon could observe the intermediate results and deal with any complications, would probably be needed to achieve the desired nose shape. A procedure like the one depicted in the show, involving a vial of nanomachines injected into your arm, and then them migrating through your body on their own to a specific place where they alter your tissue as you scream in pain and watch your appearance change in a matter of seconds, will never be a reality.

Episode 13

There will be invisible force fields. In this episode, the police go to speak with an imprisoned man, and we see that good old fashioned steel bars have been replaced with invisible force fields. This is another ubiquitous sci fi trope that makes no sense. There is no force that we could harness through any type of technology that would block physical objects in the way that fictional force fields do. The only device that can approximate its effects is a “plasma window,” which is comprised of a flat plane electromagnetic field that is pumped full of super hot charged particles. It would burn any person or thing that passed through it, though it wouldn’t physically “push back” against them. If you had a running start and were willing to suffer injuries, you could get through one.

While it’s likely that plasma window technology will get cheaper and better, the fact that they require large amounts of power and injure anyone who touches them will curtail their use. In 2048 and beyond, jails will have metal bars like they do now.

3D bioprinters will be able to make whole human bodies. This episode’s villain is another disturbed criminal with access to advanced technology. He kidnaps people and takes them back to his lab for illegal medical experiments that last for days or weeks. To cover up their disappearances, he uses a large 3D bioprinter and their DNA to make dead, whole-body copies of them and then dumps the manufactured corpses in public places at night. The discoveries of the fake corpses are meant to lead the police astray, since they’ll never assume the victims are actually alive and being experimented on.

Ultimately, it will be possible to “manufacture” whole adult human bodies in labs (Blade Runner’s Replicants were examples of this), though 2048 will be way too early. By then, the best that 3D bioprinters and related technologies will probably be able to muster is manufacturing some types of tissue (skin, cartilage) and simple organs like bladders and tracheas. We can technically already do this, but the results are usually of poor quality.

Links:

1. The “Cardiocam” mirror can measure the heartrate of anyone who stands in front of it. https://www.livescience.com/15469-cardiocam-mirror-mit-siggraph.html
2. Cheap, tiny cameras like the Himax HM01B0 are already available. The technology will be even better by 2048.
   https://petewarden.com/2022/06/15/why-cameras-are-soon-going-to-be-everywhere/
3. Research is ongoing on drug/therapy combinations that could fade traumatic memories away.
   https://www.medicalnewstoday.com/articles/a-mind-without-fear-could-this-potential-drug-treat-ptsd-and-anxiety#New-drugs-necessary