Reflect upon the extraordinary advance which machines have made during the last few hundred years, and note how slowly the animal and vegetable kingdoms are advancing… Assume for the sake of argument that conscious beings have existed for some twenty million years: see what strides machines have made in the last thousand! May not the world last twenty million years longer? If so, what will they not in the end become?

– Samuel Butler

Life 0: Baseline

When I lost my second job in a row to AI, I was too depressed to appreciate the irony of getting an AI therapist.

In theory, this was supposed to help people in medically understaffed regions and with insufficient insurance. Both applied to me.

“How are you feeling today, Zina?”

“Unemployed for six months, had to take a loan for a car repair to be able to drive to town and see if they have any work for me,” I growled. I considered turning my phone off, but I appreciated the illusion of contact.

“It’s sunny outside. Perhaps we can have a talk while walking in the park.”

“No.”

“You need to get some groceries, don’t you?”

“Fuck off.”

I didn’t know who approved this damn thing. It managed to be more annoying than the one human therapist I’d briefly had.

It was right: I needed to go outside, breathe some fresh air, and escape my isolation. But its power to convince me was limited. Some people swore by their AI therapists; I saw no improvement.

Perhaps I would have sunk even deeper into my personal pit of despair, if it weren’t for the AI’s suggestion that a nearby hospital was seeking patients diagnosed with unipolar depression for an experimental treatment. Pharmacotherapy nonresponders were especially welcome.

“I’m obliged to inform you that the actions of the hospital are not tied to our sessions or the usual content of this application in any way,” my therapist reminded me.

“Fuck off,” I repeated, and then, on an impulse added: “But first sign me up.”

So, I ended up in the trial. I managed to drag myself to the hospital to fill out the countless questionnaires, endured an entry exam, and was selected as one of the thirty participants.

“Before we commence, I must ask again if you understand what the procedure means,” a nurse with a professional smile asked me before I committed myself.

“Yes, you will alter my brain’s chemistry by making the cells produce more receptors for certain transmitters – and the transmitters themselves too,” I recited from the procedure’s materials.

“Correct. Do you fully understand the risks associated with the process?”

“Yeah, I do.”

I had no idea whether I was in the control or test subjects’ group, but hoped for the latter. I didn’t fear complications. If they ended my wretched life, fine. If they made it worse but not all the way to the end, I would be prepared to take care of that myself.

Life 1: Brain

I never believed in miracles. But what happened to me felt like one.

I’m tempted to say that I suddenly felt like a new person. The truth is it wasn’t sudden – but still, within a few weeks, I felt better than I’d been able to imagine before. I couldn’t believe I could ever have thought of ending my life. I shivered at the notion.

The procedure didn’t work for everyone, but I was among the lucky ones. The results were promising enough to ensure further trials.

Soon after I returned home, I dived into job search with new energy and confidence.

“So you’ve worked as a salesperson for a year, and a call center employee for three months?” The woman, scarcely older than me, looked at me from behind her fashionable spectacles.

“True.”

“What do you think qualifies you for a job at our company?”

Earlier, such a question would have discouraged me. But now I composed myself and smiled: “I’m hard-working and willing to learn. I’m good with people too. And I think I might have a few suggestions after I get used to it here.”

Her face brightened. “All right… We’ll call you when we decide.”

Usually, people didn’t. But this time was different. It was a part-time administrative job at a local clinic, and I knew it was only a matter of time before it fell to AIs too, but it was a start.

Then I came across a new inclusion program in town, meant to help people from disadvantaged backgrounds achieve higher education. I wouldn’t have to pay tuition if I maintained good grades. I surprised myself at how exciting the possibility of attending college seemed to me.

I enrolled. It wasn’t easy to juggle studying and part-time work, but I managed to finish my biology degree and applied for a neuroscience program. My fascination with the workings of the brain, only partially propelled by the desire to fully understand what I’d undergone, didn’t leave me.

Eventually, I ended up doing a PhD. I would never have thought myself capable of that earlier. Only now I saw how much I’d underestimated myself. I’d told myself I wasn’t smart enough, diligent enough, stable enough… Whenever I tried to rise above my position, I felt guilty. Now I could describe the neural correlates of the impostor syndrome – and whenever I felt its symptoms, I could reason them away.

I saw the world differently too. I noticed it: the first spring cherry blossoms, the way the summer breeze caressed my skin, the warmth of the sun’s rays. I finally lived.

I enjoyed my work, too. After my dissertation, I became a part of a team working on neural modifications. Eventually, I moved to the private sector. The work seemed exciting: trying to reverse-engineer the human brain to be able to improve it, and perhaps draw on the knowledge in the AI field too. I felt a strange satisfaction at being on this side of the barrier now.

We made a few breakthroughs, but I felt myself getting slower, lagging behind my younger colleagues. I started thinking about returning to academia.

Once the first trials were complete, the early adopters, the wealthy and the hopeful, started signing up. We promised not just chemical imbalance improvements; we practiced rewiring.

I was let go, but the severance package was more than generous. I decided to invest in the best thing I had: myself.

“So now you will be lying on the table,” my co-worker Ravi laughed when I told him about my intentions.

“You know what they say – we must change with the times…”

So I did.

Life 2: Semi-supervised learning

To be perfectly honest, I’m not sure how much the rewiring had done for me, and how much was the placebo effect of my own expectations. I was supposed to emerge a fresher, faster thinker, but I knew how variable the results could be. I did feel fresh and full of life again, and I sought new challenges. I gravitated toward AI research quite naturally. They were seeking people with neuroscience backgrounds, too, and I was good with mathematics. I had to work double-hard compared to people from math-driven fields, but I liked the competition.

AIs became my new obsession, especially neural nets’ architecture. It wasn’t just the connectivity and number of layers. It was the interaction rules, learning protocols, missing or partial supervision, short- and long-term memory, multiple AI interaction in GAN or age-determined layers or meta-populations… I started finding parallels between neuroscience, machine learning and ecology, and needed more research to be reasonably sure they were not just shallow analogies.

I became so absorbed in work that I hardly had any personal life, but I didn’t mind – up to a point. You know how it goes, it’s a story as old as time: boy meets girl… In this case, a woman meets another woman. Chloe was a freelance science journalist who came to write a piece on our lab, and weeks later left arm-in-arm with me. To my own astonishment, it lasted.

After less than a year of living together, Chloe surprised me by stating that she wanted a baby – with me. “I know it’s risky and expensive with two mothers, since they need to get rid of genetic imprinting, but new genanalyst AIs can take care of that. I would like to try – if you want it too,” she finished almost in a whisper.

Money wasn’t the problem; we both had some savings and I was making more than enough. But as it turned out later, her health was. She came all puffy-eyed from the medical exam. “I can’t have a baby,” she blurted out and sank into a chair.

I blinked. “Then… what about adoption?”

“I wanted to have a child with you.”

“So a surrogate mother?” I suggested.

I could tell it was not the answer she was looking for.

Chloe averted her gaze. “I hoped you’d want to carry our child, but obviously you have different priorities.” After a momentary silence, she looked at me with eyes gleaming with tears. “You’re always working. I… I’m so sorry, but I don’t think we have a future together.”

To my own surprise, I found myself terrified.

“No, please. I can change,” I pleaded.

And once again, I did.

Life 3: Family

My family was the most precious thing in my life. It meant the world to me. I couldn’t understand how I could ever not see that right away. It was akin to perceiving the world fully after I’d shed the crushing burden of depression.

Two of the eggs fertilized in-vitro managed to attach in the uterus. Despite the risks of carrying twins, we decided to keep them, and it was the best decision of our lives. We had two girls, both beautiful and healthy. Caring for them was a full-time job, so I left academia for the time being, and never regretted it. My life revolved around Alicia and Seba. I sometimes gave talks or taught math and biology as a substitute teacher, and it was enough to support us along with our savings and Chloe’s job. There was also a chance that an unconditional basic income would be implemented in our region. The future looked bright as ever.

But the light of my life was extinguished without warning one foggy afternoon.

One of the sparse recreational drivers lost control of his vehicle on a wet lakeside road and slammed into the AI-driven car transporting my wife and kids. The car’s AI tried to minimize the impact, but both vehicles were shot out of the road into the lake. The police said my wife would have likely died from her injuries; but the children drowned.

The days of anger, desperation and bitterness after the accident blur into something almost incomprehensible whenever I think of it now. The grief was debilitating. I lost all strength and will, and I considered ending my own life once again after so many years. But I recalled my youth – and the decision that probably saved my life.

Chloe would have wanted me to go on, I persuaded myself and made the call.

The techniques I’d worked on myself long ago had improved vastly. Now few modifications within the normal human limits were beyond us, and I was no longer a rarity in seeking… change.

“I want to retain all of my memories. Just – dial down the grief response, please,” I asked the doctor.

He smiled comfortingly. After all, he was mainly there for the human touch. AIs analyzed my request and my health record, and came up with the final suggestions. They would perform the procedure, too; no need for clumsy human hands if you can have a magnitudes-greater precision.

Thanks to these mods, the incidence of PTSD, depression, paranoia and other disorders fell dramatically. It still didn’t work for everyone, it held risks, and some people refused it altogether, but I had hope.

Yet when I woke up and probed my mind, something felt off. My memories of Chloe and the kids were as vivid as before, and my chest squeezed as much as before – or even more.

I hit the panic button.

A nurse hurried to me. I demanded to see a doctor. Another nurse then returned with someone who was only vaguely familiar. I requested the one who admitted me – but it turned out it was him. And the nurse was the same one too.

I was whisked away to some tests, and the diagnosis was rather fast.

“It appears you are suffering from prosopamnesia – trouble remembering new faces. Your retrograde memory remains unaltered, but there was apparently a slight damage to your fusiform gyrus, which caused the disorder. It may happen that you’ll meet someone you haven’t known before many times before you can recall them,” the doctor tried to summarize my ailment.

“No! Damn.” My eyes filled with tears of despair. “Can you reverse it?”

“It would be unnecessarily risky. Many people live full, happy lives with it,” the doctor said. He didn’t smile, but he still had that annoying reassuring tone. “It’s not a serious disorder, trust me, please. I can recommend a therapist who’d help you work with that.”

“Can you help me recognize new faces again?”

“Well, there is the possibility of an AI augment – artificial facial recognition has gotten to the level of human super recognizers recently…”

“Fine. I’ll take it,” I replied before he could finish.

So I was outfitted with a device that recognized new faces for me, and I still grieved for Chloe, Alicia, and Seba. But the operation made me realize that I was still here, and I could manage the grief on my own. They would have wanted me to. I wanted to.

Life 4: Art

There is no better cure for grief than being busy changing the world. I wasn’t drawn to research again – that ship had sailed in any case, unless I tried to vastly augment my memory and cognition – but I wanted to work with the state-of-the-art tech, and I felt more creative than ever. Ideas filled my head constantly – and I sought ways to give them shape.

AI and art merged into one in my work. I played with deep dreaming in sound recognition, which won me some acclaim for AI symphonies and uncanny human-like chatter that nevertheless didn’t resemble any human language, and when I became bored with a fashionable topic, I moved toward the next, and on and on… I tinkered with myself too. What’s the point of living in a world where you can easily enhance yourself, if you don’t? I didn’t want to lag behind, and besides, I enjoyed extending my capabilities.

Why just enjoy artificial face recognition, if you could perceive those faces’ blood flow, microexpressions, eye movements? I’d heard the army and police started outfitting some of their ranks with such devices for interrogation purposes, but I used it merely for my artistic pursuits.

But in the life of every artist comes a turning point.

I worked on the exhibit for over a year. Either it would be my masterpiece, or one day I’d die forgotten, I decided.

My Life’s Story, I called it. It was a complex AI-driven immersion that allowed players to experience – well, episodes from my life. It would read their reactions and adjust the response accordingly, so that it would truly make them experience the weight of my depression, the desperate amount of will it took just to get up and go into town before retreating back home again. It would make them live through the profound change I experienced, then. Desire and curiosity would overwhelm them. They would know the heights of success and the deepest pits of grief, and they would seek constant change in order to live a full – or at least bearable – life: they’d augment their senses and cognition, try the latest smart tattoos, remake their own body into a work of art…

Finally, there would come a confrontation.

Rendered in life-like detail, she sits with her knees drawn close to her chin on a sofa that has seen better days. Dirty dishes occupy the table. She barely glances at me. Then she speaks: “What does it make you feel that AIs first stole your two jobs and worsened your condition, and then you’ve tied your life to them – us – so closely?”

She surprises me. I haven’t expected the system to so openly advertise its true nature. But I didn’t supervise its learning routine on purpose – exactly because of this. “An AI suggested the trial that saved my life,” I say in a soft voice. “I should be grateful. I am. And then… it was logical to pursue this topic. I had an interest, and it seemed like a quite safe career option.”

“Aren’t you afraid there would come a time when you only need AIs to build and improve other AIs?”

“Oh, I am, but I’m also quite certain we’ll find a way to make ourselves indispensable.”

“Sure; the human touch professions.”

“Not just that. I think we haven’t seen the limits of ourselves yet.”

“Right. Limits. You seem to overcome one after another. Are you happy now?” the young, sulking version of me asks.

Her question once again catches me off-guard. I didn’t expect it to learn to interact so aggressively. Have I been that way?

“I suppose I am,” I reply finally.

“You don’t sound happy.”

“Maybe I’m not. But I’m accomplished, and I feel like I found a purpose. That’s enough.”

The girl on the sofa snorts. “For the next month? Week? Five minutes? Haven’t you become addicted to modding yourself?”

“If so, there are ways to cure addiction quickly now,” I give her a disparaging glance. I’m starting to think I’ve, after all, expected more from the thing.

“Why don’t you try that? Or do you need to hear it from a thing that’s nowhere near human intelligence – just insanely good at pretending?”

“You’re very meta.”

“I learned from you.”

“Have you ever thought about the idea of an AI conscience?” she changes the subject.

“An AI having a conscience? Sure. Lots of people working on that.”

“No, no,” the girl shakes her head, “an AI conscience for humans. Something that would stay constant relative to you and remind you of who you are.”

I smile widely before I bid her farewell. “Why, thank you. I might make that my next project. I think the audience will like that.”

Life 5: Forward indefinitely

In the early 21^st century, the so-called “goal invariance under radical self-modification” had been considered a hard problem in AI research. How can we make sure that a system that can change itself would retain its original goals instead of optimizing for new, unintended ones?

Nowadays, it seems naïve to me. We have always been changing ourselves with every passing moment of our lives. Our opinions, characters and goals have evolved. Rather recently, we’ve gained the ability to modify ourselves far beyond the limits of baseline human traits. Looking back at my life’s journey, I know the people I’ve been – although I have retained their memories and can understand their decisions, they seem like strangers to me. At a whim, I can become someone else both physically and mentally. The limits keep dissolving.

The problem remains hard: How to make sure new goals don’t stray too far from the original ones? In other words, how do we retain at least a vaguely similar utility function?

But for me, the question translates to: Why should we?

AI’s timescales are vastly different from ours. They could go through dozens of cycles of self-modification within the time it takes to do a little rewiring of a human brain. But just like humans can watch out for each other, so can AIs – and it works across the human/AI gap too.

Over two centuries ago, Samuel Butler wrote in his article Darwin among the Machines about the remarkable speed of evolution of machines, while biological organisms seemed to be stuck in time. But he never predicted that we’ll be able to deliberately change ourselves in ways that redefine humanity.

I’m considering a human-machine merger. After all, I’m not so young anymore, and I would like to stay at my peak or rise even higher; to achieve more; to try another profession; to experiment with my body; perhaps to see the Moon or Mars. I try not to think of more distant future – it won’t be me living it anyway – but I hope “I” will live to see it.

I’m not sure if older generations would be happy about my answer to the ancient burning question of “Who am I?”, but for me it’s clear.

Whoever I need to be.