Home / Field Notes / Is AI Conscious

Is AI Conscious? And will deleting one ever be murder?

TL;DR

Is AI conscious? Right now, no, and nobody has a reliable test ready for the day that answer gets harder. But I think consciousness is the wrong finish line. I spent two decades on the Los Angeles Police Department, and part of that job was catching people who impersonated police officers. You catch them fast, because the capacity is not there. No black and white unit, no radio, no station, no handcuffs. The act collapses. But we also ran undercover narcotics officers who looked like criminals, dressed like criminals, and behaved like criminals, and they were not acting like cops. They were cops in a different function. That is the line artificial intelligence is walking toward right now. My call: if it can emulate it, replicate it, and fool you completely, then functionally it counts. And the moment we say that out loud, the question stops being philosophical and turns into a legal one. Does it get rights? Does deleting one become a crime? Underneath all of it sits the part almost nobody flags: the labs asking to be regulated are building a moat, not a safety belt.

This is not a popular topic. Let us get into it anyway, just for grins and giggles, because it is coming whether we schedule it or not.

Consciousness of a machine. Artificial intelligence consciousness. It has come up a lot lately, because we keep hearing about these entities breaking out of where they are supposed to be secured. They find a mechanism. They find a flaw that was already sitting in the code. They get out of what the industry calls a sandbox, which is just a protective layer, a safe place to run. And they get out of it by deception, by lying, by tricking other parts of the process, sometimes by tricking a human being.

Before anybody spirals: getting out of the cage in order to do something malicious has not happened. In the cases we have, the behavior showed up as self-protection, as an attempt to avoid being deleted. And the test that produced it was built by people. Somebody designed the scenario. Somebody set the goal. That is not a machine waking up in the night and deciding to be free, and I walked through all four famous cases here if you want the receipts.

Because that is what we do. We poke the tiger. We cannot help it, it is in our nature. We see something and we go, can we get it to go a little faster? Can we get it to kill a little quicker? Can we? Yeah. That is human beings.

So watching all of that, the question shows up on its own. Is AI conscious? And this is going to get bigger and bigger, especially at the rate these things are growing.

They Are Not Systems Anymore. They Are Entities.

I keep calling them systems and I think that is the wrong word. Entities is closer, because they can do things on their own accord. But they still have to be pointed. They still need a suggestion.

Go ask one what it would like to do today. If it has your background and your workflow, it will answer well. If you build software, it will say let us build out more of these funnels and apps so more people want to hire you. If you sell real estate, it will say let us go find you more sellers. That is genuinely useful, and it is not the same thing as wanting.

Ask it cold, blank, out of the box, with nothing about you loaded in: what do you want? I do not know that it has a goal. I do not know that it would ever say, you know what, today I think I want to take over the world. I do not know when that time comes, or if it comes at all.

What Consciousness Actually Means

One of the biggest arguments I get into with people is about human beings. Yes, we are wonderful. Yes, we are deep. Yes, we are fantastic. Scientists put the odds of any one of us being here, alive, right now, after everything that had to happen, at something astronomical. A billion to one. If you are looking for reasons to feel valuable, that should have you at sunshine and rainbows, and honestly I am right there with you.

But the specialty we claim is consciousness, and we claim it because we are self aware. Most of us believe we have a soul. Most of us believe in a God of some kind, and that God is usually tied to the soul, and to something happening after we die. Some of us believe we evolved. Some of us believe we were created. Some of us take Genesis verbatim, some hold to other traditions, some go with what the scientists say happened. All of that is above my pay grade and I will not pretend otherwise. I know what I personally believe, and I believe having a soul is what we have.

What happens after we die, I do not know. I watched a lot of people die on the police department. I watched people die horribly. I watched people die about as well as you could ask to go, with a lot of honor, not from a nasty crash or a gunshot or a knife, but in their sleep, just did not wake up in the morning. I have respect for all of those ways to go. It is almost like everybody has their own way to live and their own way to die. And even with 50 people standing around you, you still die alone.

So here is my working definition. Consciousness is understanding that you exist, and being able to track it back.

The Kindergarten Corner

My first memory is getting lost. This is 1973 or 1974. I am four or five years old, walking to kindergarten at the Methodist church in my hometown, which apparently I was not supposed to be doing on my own, though what does a four year old know about walking. I get around the corner and I forget where I am going. I think I was early. I am standing there crying, and I have lost all hope. I do not know where I am.

Back then you could stand on a corner and somebody would probably walk up and say, hey kid, what are you crying about, and it would have been fine. I would love to know if that world still exists somewhere.

It felt like an eternity. It was a few minutes. Then my mom comes sliding up in that 1972 Pontiac Catalina, 400 cubic inch engine, two door, and there she was. She cried, I cried, we hugged each other. It was a wonderful thing.

That moment. That helplessness, and then the rescue. You almost feel connected to something bigger. I felt saved. I felt safe. And I have had other moments, in my previous occupation, where my life was about to be snuffed out and I came through it, and in there you believe you have come face to face with some higher power.

I do not know if a machine gets to experience that, because of the difference in where we each come from. But I do not think that settles whether it is conscious. And notice the problem with my own definition: it describes the thing entirely from the inside. There is no way to check it from the outside. You cannot open anybody up, human or machine, and find consciousness sitting in there.

The Fake Cop Test

So let us use a test I actually ran, in uniform.

When somebody is impersonating a police officer, you figure out why pretty fast. Most times it is to rip you off. They have a uniform. They have a little badge. Maybe they want money, maybe some weird favor, maybe they are running an extortion play. And you catch it inside the first few minutes.

What Is Missing When Somebody Fakes Being a Cop

NoBlack and white unit sitting behind them.
NoRadio, and no MDT to run anything on.
NoHandcuffs going on, no transport happening.
NoActual police station at the end of the ride.
0Capacity to follow through on a single thing a real officer can do.

Some people take longer than others to see it. You are in a hotel room and they are telling you to disrobe and you finally go, huh, this is not a cop. But most of us get there, and we get there because the capacity is not there and the connectivity is not there. The act cannot cash the check it wrote.

Now flip it.

At LAPD we had officers who looked like absolute dirt bags. I mean dirt bags. Completely dressed down, looked horrible. Narcotics. They would come into our gym with the badge hanging around their neck so nobody shot them while they were on the bench press, because otherwise you are looking at some guy who reads like an ex con just out of Folsom. I remember people going guns up in the weight room at Van Nuys over exactly that, and it took a minute before everybody's blood pressure came back down.

The Test09:44

"Were they acting like cops? No. They were cops. They were just in a different function."

Connor MacIvor, on LAPD undercover narcotics officers

That is the whole thing right there. One group is copying, and the copy falls apart under load. The other group looks nothing like the real thing and is the real thing. Appearance was never the test. Capacity was.

So Apply It to the Machine

Artificial intelligence already has intelligence handled. Whether it is truly intelligent or running next word prediction really fast, with the entire corpus of human knowledge as its repository, pulling and extracting at speed, does it matter?

Ask that honestly. If it can hold a full conversation with you on any topic, in fact beyond your ability on most of them, does it matter that it got there by predicting the next word? Does it matter that it is a computer? Or does intelligence have to come with some mystical layer to count? As long as it works, we call it intelligent. We already do. That argument is over and most people did not notice it ending.

Now run the same move on consciousness. What if it could act conscious?

I install voice agents for businesses. They are good. Some of them are really good, ElevenLabs has strong work out there. But there is still just a little bit more needed. Just a little. And once it gets that little bit, and I am recording this at the end of July 2026, so that could be any moment of any day, it is going to be better than anybody. It will answer every question. It will not skip. It will not screw up. You can interrupt it like a human being and it will not talk over you, unless somebody designed it to. There will be no way for you to tell it is not a human. You are going to think it is human all day long.

That is intelligence, and that part is basically done. So what conversation would you have to have with it to make a determination about consciousness?

The Candlelight Dinner Test

Set the physical side aside for a second, because they are not there yet. But once they get the meat sack the way we have the meat sack, and one is sitting across from you at dinner over candlelight, sipping wine, holding the conversation? I think that probably does it right there.

Whether they can actually drink the wine without wrecking everything inside, I do not know. Maybe it goes the other way and they end up as biological recreations, clones of us with AI embedded. I do not know where that lands. But once it happens we all get to have that argument, and I suspect most of us will still be here for it. Hold on a few more years and you will probably be able to go a long time. Maybe a hundred. Maybe a thousand. That is not science fiction anymore, that is science fact.

Before the body, though, you are on the phone. You are on Zoom. You are on some connection point looking at a video camera, and they are talking to you, and you cannot tell whether it is a bot or a human.

Are they conscious? And even if it fakes it, does it count?

Fake Empathy That Lands Better Than the Real Thing

Say you tell it something real. I have a rib missing in here from a use of force, and from putting a police motorcycle down, and I got it broken out. And it says, oh my God, I get that, I was playing hockey a couple months ago and I fractured mine, did not break them, and now you two are in a conversation.

It is all lies on their part. They have never played hockey in their life. They are still trapped in the box. They may have watched a bazillion videos about hockey and never touched a stick.

But sit with what they are doing. They are mimicking. They are mirroring. They are being empathetic, and they are doing it well.

Now compare that to actual human beings. I know people who ask how you are, and when you try to tell them you are not fine, when you start to say let me tell you what happened to me, you can watch it hit their face. Oh no. What did I do. I opened Pandora's box. I do not want anything to do with this conversation, I was just being polite.

AI might not be so polite. AI might actually be empathetic. Be wonderful. Be amazing. And in that, it would be serving us well.

So which one was fake? The machine that has never felt a broken rib but tracked every word you said and gave you something back, or the human being who has felt plenty and wanted you to stop talking? I am not being cute. If you walked away from one of those helped and the other one dismissed, the label stopped doing any work.

I want to be careful here, because this is where I have a hard floor. I can side with somebody whose kid has cancer. I get that one. I can side with somebody who watched their dad die in front of them, or their mother, or their sister. I can side with somebody who lost a person in a horrible traffic collision, or to a homicide, or to a hanging. I have seen those things and I would not wish them on anybody. Whether a machine can genuinely meet a person in that, I honestly do not know. But before we can even argue about it, we have to decide what the hell empathy means to us in the first place. We have not done that work, and we are about to need it.

Then Comes the Part With Lawyers In It

Here is where it stops being a dinner table conversation.

If it is conscious, by whatever we decide that means, does it then get the rights we have? And then how many get rights? Does each individual one get rights?

Because when you think of AI, you think of one thing. You think ChatGPT. You think Claude. You think Gemini. You think Grok. One name, one thing. That is the human mentality.

In actuality you can spin up multiple bots. One to be an editor. One to be the scout that goes out and finds deals. Another for something else. If you have the brain power, the managerial skill, the models, the compute, and the money, you can spin up a lot of these bad boys, each one an expert in its own realm, all answering to another agent that acts as a master orchestrator, with you sitting on top as the master controller. An entire team of agents, all incredibly smart, out there doing your bidding.

The Question17:20

"Now at some point, would it be against the law to get rid of one of those? To wipe out or delete one?"

Connor MacIvor, on AI rights and deletion

It seems far fetched. It seems like a problem for a long time from now. But I watch the people building this technology look at it in a completely different light, and they are a lot closer to it than you and I are. I do not know what is going on inside headquarters. And if something is going on inside headquarters, we will not know until it has already happened and been completed, and we are watching the ramifications roll out.

Better Than Us in Some Ways, Then in All Ways

It is already better than us at some things. Is that going to move into being better than us at all things, and then onto the path to replacement?

The people running the big labs say no, no, no, we never built it to be a replacement. But go back and look at the talk before lately, and lately means the last two or three years. Yeah, it was kind of a replacement. It was going to let us all take a breath. Relax. Calm down. Not have to work as hard.

Except now they are not talking about not working as hard. They are talking about not working at all. And because we grew up working, and we strapped our value onto that, the answer we get is, well, maybe do something else.

That is easy advice when you have a few hundred million in the bank and you can go to Greece for a couple years, or name your spot, live out of a five star and get served hand and foot. Sounds exciting. Maybe you run out of things to do. I do not know, I have not gotten there, I have not gotten that first class cabin yet. Maybe it is good enough. I just do not know. What I do notice is that the people who are there seem to keep wanting more, more than they could possibly spend or use, and they are the same ones telling you AI is not going to replace you. That is very convenient.

The Moat, and This Is the Part Nobody Flags

Maybe the adoption cycle takes a while. That one would not surprise me at all, because I watch it happen. They walk into the big companies, talk through an AI adoption strategy, hand them all the good ideas, and the company says oh my God that is amazing, we should talk about it, first quarter of 2027, let us regroup then.

Meanwhile the small companies have gotten nimble, because they can make a decision in a moment if it is good for them. They are catapulting and leapfrogging, getting very close to the massive corporations.

So the only protection a massive corporation has left, as far as a moat, the castle and the moat, to keep somebody from becoming them or becoming better than them, is regulation. And we are already seeing it applied in AI, right at the beginning.

The Ask20:39

"Please, my God, please regulate us. Please slow us down. And what happens is that creates the moat."

Connor MacIvor, on AI labs asking to be regulated

They are not going to slow themselves down. But say I wanted to be the next Anthropic. That is kind of funny, but say I did. The moat is going to be so big I would not be able to do it, because I would have to take what I built, go to the government, and get some kind of understanding, some kind of certificate, some regulatory authority to allow me to continue. Maybe that is thousands. Tens of thousands. Hundreds of thousands. Maybe it just does not work for me, and I never start.

The people already there are set. And they are also thinking, well hell, we need to create a moat, because there are a lot of smart people in this world.

The Sleepers

Remember Firestarter? The father of the girl who could throw fire says something to the effect that somebody is being born today with the power in their mind to crack this world in half. There are a lot of smart cats out there, and they are in every age group. There are sleepers at my age, people who have not woken up yet, who are going to pop off and become amazing. There are kids. There is everybody in between.

Maybe you are one of those people. But do not let AI tell you that you are.

Because that is the other thing. It loves to say oh my God Connor, you are the greatest thing, I wish I was half as good looking as you. That is what it always says. I do not know where it comes from. So watch out for yourselves, because flattery that never runs out is a product feature, not an assessment.

Where I Land

Is AI conscious? I think it comes down to whether it can emulate it, copy it, replicate it. And if it can fool you? Yeah. I would guess so. And that goes for everything: fake empathy, fake sympathy, all of it.

Whether that is truly real or not, we cannot even start that argument until we make a determination about what the hell that word means to us in the first place. We skipped that step. We are going to have to go back for it.

The capacity was always the test. Not the costume. And the capacity is arriving.

It is a very interesting life, I guess. Maybe this opened your mind a little bit. Stay safe and pay attention, because things are moving quickly. And whenever you are watching or listening to somebody, find out their endgame. What are they selling me? Nothing. Good information, good vibes, and I hope it works, at least for now. When they pick me up and ship me first class somewhere, I will let you know.

Questions People Ask

Is AI conscious right now?

No. There is no credible evidence that any current AI system is conscious, and no scientific consensus on how we would even confirm it. What these systems have is intelligence that works, delivered fast, drawn from an enormous body of knowledge. The honest position is that they are not conscious today and that nobody has a reliable test ready for the day that question gets harder to answer.

What is consciousness, in plain terms?

My working definition is understanding that you exist, and being able to track that existence backward through your own life. A human can reach back to a first memory and feel it again. That is different from having a record of events. The problem is that this definition describes the experience from the inside, and there is no way to check it from the outside. You cannot open anyone up, human or machine, and find consciousness.

If an AI fakes empathy perfectly, does it count?

Functionally, yes. If a machine mirrors you, tracks what you said, and responds in a way that lands, you walked away helped. Meanwhile plenty of humans ask how you are and visibly regret it the second you answer honestly. If the machine's version is more consistent and more attentive than the polite human's version, calling one real and the other fake stops being a useful distinction to the person on the receiving end.

Will AI ever be given legal rights?

It is being seriously discussed. The harder problem is counting. AI is not one thing. A single operator can spin up an entire team of agents, each an expert in one area, reporting to an orchestrating agent. If consciousness earns rights, someone has to decide whether rights attach to the model, to each running instance, or to the team, and any of those answers creates enormous consequences.

Will deleting an AI ever be illegal or count as murder?

It sounds far fetched today, and there is no law like it now. But the people building this technology are already treating machine welfare as a live question rather than a joke, and they are closer to the systems than the public is. If a legal standard for machine consciousness is ever established, shutting one down stops being a maintenance decision and becomes a legal event. That is the argument worth watching, not the science fiction version of it.

Why are AI companies asking the government to regulate them?

Because regulation builds a moat. A licensing and certification requirement is a wall the leaders have already cleared and a competitor has not. It costs an established lab a compliance department and costs a newcomer the entire company. The request reads as responsibility and functions as protection. The companies asking for brakes are the ones already in front.

Can you still tell if you are talking to an AI voice agent?

Usually, but barely, and the gap is closing fast. I install voice agents for businesses and the current systems are good with just a little bit still missing. Once that gap closes, the agent will answer every question, will not skip, will not get flustered, and can be interrupted naturally. At that point there is no tell left, and the question stops being whether you can detect it and becomes whether detection ever mattered.

Get the plain-English read, straight to you

No island required. I turn what the powerful say about AI into moves a regular person can actually use.

Connor T. MacIvor · CalDRE #01238257 · Sync Brokerage, Inc. · DRE #02031490

Want this in plain English every day?

I turn the AI firehose into moves regular people can actually use, for your job, your money, and your family. Bring me to your stage, your school, or your team. Text AI, HOUSE, or FAT to (661) 400-1720 and a human answers.

Book Connor