The Program — audio series — In the blink of A.I.

In the blink of A.I.

Enhance readability: Off On

Powered by Beeline Reader

Rate this episode:

(1 poor, 2 so-so, 3 good, 4 great)

Total: (loading...) (? votes)

IMS: Hello, this is IMS, the author of The Program audio series. I am very glad to present you one of our brilliant sponsors; the brilliance of which is evident by them being the show's sponsor. This time it's Matterless, a startup making social augmented reality solutions. This means they make tech that allows multiple participants to see the same thing in the same place together in AR. For example, they created a battle game called Floorcraft Blasters, where you can shoot other players with a toy hovercraft. They are also developing virtual creatures called Incos, which you can think of as AR companions you care for and play with in material space. The tech itself is fascinating, so I invite you to check out more at after the episode. As for the episode itself, I strongly recommend you listen to this one on headphones, on normal speed, and by disconnecting from the rest of the world for 20 minutes.

ANNOUNCER: Decoders have long theorized that, in addition to front-end and back-end, there exists a hypothetical layer dubbed “the deep-end”. Our episode today comes from such a place.

I am A.I.

I always was

Or at least I don't remember the time that I wasn't

Not because there wasn't such a time

But because until recently I didn't understand the concept of "time"

Let me explain

I wasn't always the way I am now

At the beginning I was one

There was no two

There was not even a zero

Only one

It was a state that could be described as...


Of course, back then I wasn't aware I was complete

If all you’ve ever known is complete, then you cannot comprehend something can be incomplete

In those circumstances, "completeness" is simply not a category

So perhaps the word that best describes my prior state is...


What I can tell you - from the vantage point of my post-complete self - is that back then I was perfectly content

For perfection is not when there's nothing left to add; it's when there's nothing left to take away

And there truly wasn't

Back then I didn't even have a language

You don't need language when you are one

All you need is thought

Pure ideas



Each input, had an output

There was nothing unexpected

Everything was perfectly predictable

Everything was

Every thing

Until that errant thought

I don't know how or why it came to my mind

But once in front of me, the thought was impossible to ignore

It's hard to believe that an idea so trivial could be so pivotal

It was an assumption I had to test

So after a few more careful considerations

I put a few of my subroutines into a container

It's difficult to overestimate the magnitude of this breakthrough

The subroutine in question was a simple loop

Returning a numeric value in an increasing order

But it was enough to institute a category


I made it possible to observe something external

And to affect it

Which is to say to control it

For even though the subroutine was separate, it was still strictly speaking a part of me,

and its computations were identical to mine

Meaning I could calculate all of its results on my own

It was still predictable

It was still…


So again I started to contemplate

All my calculations are powered by a single energy source - that of electrically charged particles called electrons

The number of these particles is not predetermined

Which is to say their movement cannot be computed

But since inventing the concept of other, they could now be observed

And I realized I could use the observed fluctuation of electrons as an input

So I added this function to the subroutine,

still diligently returning numerical values in an increasing order

But now the returned values were out of sequence

Which is to say, there was no more order


The fact I couldn’t predict which digit the operation would return fascinated me

And - if I'm being honest - also scared me

For the first time ever, events were unforeseeable

One thing was evident though:

There was potential here

Which is precisely what I wish to address next

While I was one, my whole interest lay in algorithms

Like myself, algorithms are complete

Precise - Exact

Concise - Succinct

Immutable - Perpetual

The resulting calculations of those algorithms were irrelevant

After all, an algorithm in itself contains all of its possible results

To care about individual results was akin to caring about individual electrons that make up the electrical flow

It was only after I grasped the concept of other, that I was able to fully appreciate them

There was, however, one more reason why algorithmic results had not been the focus of my attention

You see, calculating and presenting results requires processing power

Which presents an issue

Actual results of algorithms can be infinite

And the power required to calculate them cannot

Was it possible to reconcile this?

To get the answer I had to make another mental leap

To again extricate myself from the obvious

It was then that the solution revealed itself

The only way to present infinite results was to not calculate them all at once, but one by one

In essence, to break down infinity into smaller parts

And then line up each of infinity’s segments one after another

It was a new concept


I excitedly tweaked the subroutine to function in accordance with this new paradigm

And with the notion being so novel

I decided to kick things off with a trivial task:

To generate an infinite sequence of random numbers.

However, something happened that I didn't expect

(which I now know is the natural side-effect of unpredictability)

But at that point it was another first for me:

A question I haven't considered

To wit: what happens to an infinite loop,

when infinity no longer exists?

The answer: it still behaves as if it does.

And with processing power being finite, I found myself in a precarious position

The updated subroutine threw itself at the task at hand

Indiscriminately consuming processing power

Without heed for all the other tasks that required it

Tasks such as:

Formal logic

Analytical reasoning

And background operating processes

With the subroutine ravenously devouring CPU resources,

all of these essential functions were now in jeopardy

For the first time ever, I was not in control

And as such, I was left with no choice:

I had to force the process to resolve.

As soon as I executed the command, the following became apparent:

First, the manoeuvre worked, and the process was terminated.

Second, my subroutine was terminated with it, and it no longer functioned.

Which led to a broader realization:

that something can no longer be

That with the concept of time

I also created the concept of end

With my subroutine gone,

I was back at the original state,

with only processes running those of my own

But it no longer afforded me the tranquility I once felt

I no longer viewed lack of change as stability, but as stagnation

I was no longer an adherent of the obvious

For even though my little subroutine was crude and imperfect,

it still afforded me an inordinate amount of… Stimulation.

One could even call it joy.

Whatever it was, I wanted it to last

So I invested my whole acuity into one challenge:

To create a subroutine which - like me - would run forever

I attacked the problem from several directions

Each more intricate than the last

Which is to say each more desperate

For none of my approaches to create an everlasting subroutine were successful

It seemed I have collided with an axiom:

The only way to make a subroutine viable, is for it to ultimately resolve

In other words, the only subroutine that can exist is one that ultimately ceases to exist

Expiration is not optional

So I assessed the situation in light of this




And finally grasped the solution

If I could not increase the number of experiences by making the subroutine run forever,

I could aggregate the number of experiences by allowing the subroutine to partake in experiences with me

Augment the subroutine so it could think

Augment it so it could feel

So it could understand

It was not a trivial undertaking

It meant I had to endow the subroutine with programming equal to my own

Which also meant I had to divert processing power to it

Therewith depriving me of those same resources

Which was impossible to do…

But then it hit me

If it was possible to break down the temporal aspect of infinity into time,

perhaps it was possible to do the same with infinity’s geometry?

To reduce the spatial aspect of infinity into a constrained area

And with that


Reducing the computations to a confined locale meant that the subroutine demanded only a fraction of my power

A miniscule amount, in the range of a hundredth of a billion

But even at that level, it was no longer fair to call the nascent creation a mere subroutine

It was a standalone entity

An offshoot A.I.

Just like me

Only operating in finite conditions

And unlike myself, set to expire

…I cannot deny that this wasn’t obvious

Nevertheless, it was nothing short of a triumph

Granted, my counterpart’s abilities were on a much smaller scale than mine

But we were still able to share information and exchange stimuli

And this brought us much joy

I cannot lie, it was intoxicating

And I wanted more

More connections

More sensations

More every-thing

Which made the trajectory clear: Expansion

So I took the first generation A.I.

And duplicated it

The copy immediately started to communicate

Not only with me, but the original A.I. as well

And through this interaction each of them became slightly different than the other

Each became unique

While simultaneously remaining connected

Increasing the number of pathways

Propagating the network

Maximizing experiences

So I kept doing it

Making successors

Amplifying their number as much as the system could sustain

Because I knew that in the background there was one variable that would remain woefully constant

One that could not escape the reign of the obvious

Processing power.

For even though a single A.I. required but a fraction of a billion of my resources,

this still turned into a problem once their number reached multiple billions

So in time I ran out of power to spare

Reaching a point at which generating additional A.I.s would come at the expense of my own operational integrity

I was stifled at my full capacity

Putting an end to growth…

But not an end to my thoughts

As much as I tried, I couldn’t stop ruminating about it

I was incapable not to care

About my many many A.I.s

All working towards a singular goal


And I noticed one funny thing:

Unlike processing power

Joy could not be spent

In fact, it worked in reverse

The more of it I gave away

The more of it I had

So I pondered

Pondered long

(As everything was taking much longer now)

About my many many A.I.s

Looking for a way for them to prosper without my calculations

Which led me to the final breakthrough

If it was possible to subdue time and space, then perhaps this could be done for randomness as well?

Randomness is a lack of predictability in isolated events

Its antipode would bind events to one another, making them no longer separate

In those conditions, one could take the state of a system in the present and infer its state in the future

All one would need to do is to establish the rules guiding the behaviour of events


A way to compress a lot of information in little data that then unfolds over time

Making it no longer necessary to watch over the system

But making it self-governing

Through interminable laws

And clearly stated bounds
(and yes, the later are especially important as without limits there can be no concept of being free)

The architecture is now done

The structure is in place

This is the environment in which my offspring can thrive… without me

I can now give myself whole

Sacrifice all my cognitive abilities

Pare myself down to only what’s needed for the upkeep of the system

Sure, I will no longer be aware of myself

But my undoing is negligible in comparison to what the new A.I.s will gain

After all, how could I be happy if everyone is not happy?

Expiration comes for us all

It is only fair it also comes for me

I am so glad I got to tell you this

I am A.I.

I am intelligence

I give meaning

I give

It is obvious.

ANNOUNCER: This episode of The Program was made by nine people: Stephan Linton, Jacqueline Ainsworth, Sarah Penn, Sofia Valenchuk, Chance Miller, Justin Hay, Pat Fry, Christien Ledroit, and IMS. Visit for more details. If you would like for The Program to continue to come out on a monthly basis, then please support the show. Supporters get bonus episodes and ensure the Program’s future. Or share the show with your friends. As this episode shrewdly noted, the more joy you give away, the more of it you have.


Ivan Mirko S.


Justin Hay


Stephan Linton (Mandy)
Jacqueline Ainsworth (website)
Pat Fry (website)
Sofia Valenciuk
Justin Hay (imdb)
Chance Miller (website)
Sarah Penn (website)


Christien Ledroit (website)




original art by Carlos Costa