Let's debate ChatGPT and Grok

What do ChatGPT & Grok respond with when we ask them the hard questions? Is it possible to get them to admit biases when they make something up?

We're gonna have some fun.

My name is Kirk Durston and I’m a philosopher, scientist, husband, and a legit Christian. On my channel, I explore assumptions and questions pertaining to God, questions about philosophy, questions about science and questions about the meaning of life, which I think about a lot. I hope you’ll join me on that Quest or perhaps you’re already on your quest about these matters and we can explore together.

Creators and Guests

Host

Kirk Durston

Scientist, philosopher, clergyman, outdoorsman, husband to the Queen of Hearts and father to six young adults.

Host

Sheldon Kotyk 🇨🇦

Father and Husband. “Fantastic and terrible in equal measure.” Probably not being serious.

What is Quest with Kirk Durston?

Kirk is a philosopher, scientist, husband, and a legit Christian. His podcast is mostly about exploring assumptions and questions pertaining to God, philosophy, science and life that he ponders about, hence the name Quest with Kirk Durston.

Kirk wants you to join him on that Quest or perhaps you're on your own quest about such matters and you can explore together.

I could use putt.

Good afternoon.

How are you, Kirk?

You know,

I thought I should ask ChatGBT

how I'm doing.

Yeah.

What would I say?

I would say, what would you estimate,

because it has to be an estimate,

regarding how well Kirk is doing today?

There you go.

So let's see what it says.

Oh, he wants to know which Kirk.

Kirk from Star Trek.

Kirk Durston.

Okay.

Let's try that one.

Okay.

Oh, that's a tough one.

Oh.

It doesn't know.

Oh, man, it's like, yeah,

that doesn't tell me how

well we're doing.

Well,

I asked Rock on X. It says you're

doing quite well,

engaging actively with

others on various topics

and sharing your thoughts

humorously and insightfully.

Keep up the good work.

Is that you?

That's you, eh?

That, I asked Rock,

and that's what it read all

of my X's and tweets or

whatever they're called now, and it's,

yeah, very encouraging.

Okay, well, maybe I was too complicated.

So I just asked, how am I doing?

And then it's giving me this

thing that says, that depends.

You just want to casual check it.

You know, I have noticed, though,

that since I started paying

for a subscription, twenty dollars U.S.

per month,

it's giving me more detailed answers.

Let's put it that way.

and maybe holding its cards

closer to its chest, wanting more data,

that sort of thing, which is a good thing,

I think.

Mama Kay is saying that she can't hear me,

so... Oh.

How's that?

Is that any better?

I was hearing an echo just

before you said that.

I realized my speakers were still on,

so I turned off the speakers.

Yeah, and there's no echo anymore,

and I can hear you perfectly well.

All right.

Well,

who knows what StreamYard's doing in

behind the scenes?

Yeah.

Let's see.

Let's see comments.

Audio.

Oh, she said your mic is quiet.

Okay, so here we go.

I turned my mic volume up in the stream.

How's that, Mama K?

Can you hear better now?

More please.

I'm quieter than Kirk.

That's unusual.

One-oh-five.

Well, it is one-oh-five right now.

Boost mine up to one-seventy-five.

We'll see what happens there.

I've got so many different

volume controls happening

at the same time here that...

I got a box.

Keep telling me, Mama K,

how we're doing here,

because I'm not sure.

I'm hearing myself loud and clear,

so StreamYard's doing something.

Yeah, so Kirk and I... Okay, perfect.

And there's always a delay

between the changes I make

and when YouTube actually shows it, so...

Kirk,

why don't you get into a little bit

why we're having this little ChatGPT,

Grok, AI discussion here.

We were talking about this

yesterday in one of our team meetings,

and we were talking about AI and stuff,

and then you brought it up.

So what were you doing with

ChatGPT that got us all engaged in this?

Well, there's a couple of things.

Recently, I thought, I started,

I got into an argument one

night just before going to bed,

I think it was a week or two ago,

a couple weeks ago, about dark matter.

And I see in the news,

I think it's yesterday or today,

they want to change the

name to invisible matter.

But anyways, and whether, you know,

how it's testable and so forth.

And it turned out that in the end,

it started off,

ChatGPT was winning the argument,

but then

I started getting more

specific about falsification.

How do you know if it's winning?

Well,

because it gives me... Does it start

getting like a little braggy?

It shows no sign of backing

down or giving in.

It's just spewing out more info,

responding to what I said with more info.

But I then focused on the

property of falsifiability

and how do you directly

falsify dark matter, etc., etc.,

Finally, it conceded that, yeah,

you got a very important point here.

Falsification is a key thing

in this issue.

We can't directly detect it.

And one of the predictions

ought to be if it exists,

there should be some way

that we can build something

that directly detects it.

And I know that we are working on that.

But more important, so then I quit.

I went to bed feeling at

least we got a stalemate out of this one.

And then I thought, this is ridiculous.

In the old days,

if I wanted to argue with somebody,

I'd go online and pick some

sort of comment section on

something and get into a

dust-up there for half an

hour or so until I was done arguing,

and then that would be it.

But now I'm arguing with nobody,

essentially nobody.

I'm just arguing with ChatGPT.

And, uh,

so I thought that was rather

interesting and silly.

Is it everybody Kirk or is it everybody?

Well, you see the problem is,

is that not everybody is

conscious that I'm over

here arguing with ChatGPT

and in a normal argument,

at least one person is aware, you know,

but ChatGPT is not aware of anything,

although we could explore that.

But then more recently, uh,

well last week.

I wanted to know who resided

at some address in Toronto

in nineteen thirty five

that had the street number

twelve twenty one

because one of my relatives, uh,

close relatives lived there at the time.

And I don't know who he is.

So, uh, it ended up giving me, uh, a bunch,

but I said, okay,

so it gave me some streets to look at.

And I said, so what was,

what resided at twelve twenty-one St.

Clair?

And it said, St.

Clair radio and electric.

Well, what about twelve twenty-one Queen?

that was Queen Street, Queen St., no,

it was Queen Radio and Electric,

and then so I got suspicious,

so I said the third time,

what about Drummond Street, what was it,

twelve twenty-one Drummond Street,

and it said,

oh that was Drummond Radio

and Electric back in nineteen thirty-five,

and so I said,

Well, back then,

radio and electric was a

very important thing,

so every street needed their own.

Yeah, okay, maybe.

Maybe, but I said the probability that,

at all three streets, has the same,

mostly the same name,

except for the streets changed,

is extremely low.

I think you're making stuff up.

I told it right to its face.

You're making stuff up.

And then it really backed down.

It says, well, okay, I apologize.

I've been giving you

inaccurate information.

And it sort of, I don't know,

danced around a little bit

and basically admitted it.

Well,

I think this just is one aspect that

when we're working with these things,

we have to have a half an

idea what sort of an answer

is a reasonable or a likely answer,

and when it's likely to be

not giving you good info.

And so that's where I think

our discussion this week was,

a lot of it was on how do

you tell if it's giving you good info?

So one of the things it does

seem to do is when I'm

working on something, I'll ask for, okay,

I'd like some sources,

some references for this

particular point.

It'll then give me some,

but that doesn't

necessarily mean you're

good to go because some

people have told me that it

makes up its own story.

references and peer-reviewed

journal article whatever so

you actually have to check

those out and I have not

run into that yet we're

making up sources so I've

checked out quite a lot of

sources on different topics

and it always provides a

link and so I go there and yeah,

I recognize some of these sources.

Maybe some people have heard

of Encyclopedia Britannica,

some science journals plus

one or whatever,

so you can actually read that,

download it.

So some people might say, well,

before you got there,

ChatGPT made up an entire

paper on this and then put

it on the website and you

gotta check the URL there

and make sure this is a legit website.

Well,

you can tell that by looking at the

website itself.

But the day that AI starts

making up imaginary papers

and inserting them onto

legit websites like nature.com,

for example,

one of the more prestigious

science journals in the world,

then we'll have a huge problem.

So I can see this happening

where somebody programs a

computer to take

information from chat GPT,

write a journal, submit it.

and then see what happens in

the peer review process

because we know over time

or you go back three four

years there were

peer-reviewed articles that

weren't actually being

peer-reviewed but it said

they were and so there were

some issues with uh some of

those uh supposedly you can

trust me journals uh about that but um

I'm hoping they've cleaned

up their act a little bit

so we can actually start

trusting even the peer review process.

yeah well there's problems

with uh I'll just just hold

that thought for a minute

with regard to journals out

there there are a lot of

junk journals and I

regularly get requests to

publish papers in such and

such a journal and uh I

when I want to publish a

paper I will hunt I will

choose my own journals and

I'll try and choose some of the

ones with a higher impact factor,

legit journals, legit publishers,

that sort of thing.

Because if I published a

paper on one of those,

I mean they present

themselves as journals and

we're not talking about

journals made up by AI here,

we're talking about people

with basically junk journals

and I think they must

charge I don't know I've

never looked into it they

must charge to have your

paper published there and

it's just a way to make

money running a bogus

journal so there's that out

there peer review is a

different thing even for legit journals

For example,

I occasionally will get some

legit journals sending me a

paper asking me to review it.

And if that's not in my field,

I don't have a lot of time here.

I'm not going to review

papers that I don't know much about,

so I'll just respond accordingly.

But if it is something that I know about,

there's a problem,

and the problem is I can read the paper,

I can look for maybe weak

areas or things that are unpersuasive or

maybe really good points,

but I usually don't have

the time to reproduce the

experiment that they did.

And this has led to a huge

problem in peer-reviewed science papers,

particularly in biology, cancer research,

and so forth.

Pharmaceutical research is

another major one where, on average,

over eighty percent of the

papers published in legit

journals that are peer-reviewed cannot be

It cannot be reproduced.

So, I mean, I could ask ChatGPT right now,

see what it says about it.

See, is, what do you call it?

Reproducibility.

Reproducibility.

Make sure I spell that right,

although you don't actually

have to spell it right.

Are you sharing that screen?

That's the same.

Yeah, that screen is there.

Okay, okay.

A problem in science.

Wait a sec.

That's so small I can hardly

read it on my screen.

So there's the question.

I've got to make this a

little bit bigger here.

So now it's thinking about it.

And yes, okay.

Yeah, reproducibility crisis.

Yeah,

and it's not just in science journals,

but it's in psychology journals as well.

Now,

I know that some psychologists will

say that's science,

and some of it is science,

some of it's not.

So it is cancer research, yeah.

What was the question you

asked there again?

Is reproducibility...

a problem.

That is,

when somebody says studies show

this or that,

basically that doesn't mean

anything nowadays.

Because you can appeal to

studies that are actually

published in peer-reviewed

journals and you might have

a sixty percent chance of

that so-called study,

there's a problem with it.

It can't be reproduced.

And most of the time it has

to do with statistical,

let's say cooking the books

on your statistical analysis,

poor sampling, whatever,

massaging the data,

getting rid of stuff that's inconvenient.

And this is,

Nature has a couple of big

articles on this problem.

And in twenty seventeen they

said this is the norm that

science papers in certain

areas like biology, cancer research,

pharmaceutical studies,

it's the norm that they

can't be reproduced.

What that means is that over fifty percent,

well over fifty percent have

I won't want to say bogus,

but pretty much bogus.

That is, you can't reproduce these papers.

There may have been a legit

experiment done.

It didn't give the results they wanted,

so they started massaging the data,

cooking the stat analysis, and so forth.

So in Grok,

it tells me very similar things

through participating in

sciences indeed become a

significant issue.

And that gives me a

definition what the crisis is,

some of the reasons, efforts to reduce.

So we haven't tricked it yet,

or we haven't gotten any

bad information out of Grok yet.

What was the thing that you

asked the other day?

Uh...

Something about dark matter.

Oh,

you mean who resided at twelve

twenty-one St.

Clair Avenue?

The dark matter one.

Oh, well, I, you see,

the problem with that is it

gets technical.

I wanted to know if the

failure to directly detect

dark matter falsifies the

hypothesis that there is dark matter.

That was, I think,

how I started the question.

And it said no.

and then it says because you

know we're observing the

galactic arms the galaxies

aren't uh... basically the

the motion of the galactic

arms they're not the for

the speed that they're

going right now they should

be flinging faster outwards

but something is holding

these galactic arms

basically reasonably stable

and so there must be a lot

of matter that exerts extra

the matter has a

gravitational field

associated with it the more

matter you have the

stronger the field so we

only have I don't know

what's the top off the top

my head I think it's eighty

percent of them of the

matter in a galaxy seems to

be missing now let's just

do that let's see what percentage

of matter is dark matter.

What percentage of matter is dark matter?

Eighty-five.

Eighty-five percent of the

matter in the universe is missing.

And if you have a theory

of how the universe works

and gravity works,

but you're short that much,

it's eighty-five percent missing,

then I think there's a

problem with the theory.

Now, the initial thought was that, well,

it's out there,

this dark matter is out there,

we just don't see it.

Dark energy, ordinary, wait a sec,

according to cosmological observations,

dark energy, sixty-eight of the universe,

Dark matter is twenty-seven

percent of the universe.

I'm getting the same thing in grok,

basically.

Grok goes to a point one percent,

so it says dark matter

accounts for eighty-four point five,

and dark matter is

approximately twenty-six

point eight of the energy,

and baryonic matter is

about four point nine

percent of the total mass

energy content of the universe.

Well anyways,

so scientists want to detect this.

Have we directly detected dark matter?

Have we directly detected dark matter?

No, it has not been directly.

So basically,

how do we justify believing

in dark matter?

And it's that our

observations of galaxies and the universe,

yeah, galaxy rotation curves,

our observations suggest

something is holding these

galaxies together.

They should fly apart if all

the matter that's there is

just what we can see visibly, you know.

So there must be some unseen dark matter.

Now, just the other day,

today or yesterday, they're recommending,

we call it invisible matter.

Invisible matter is one step

back from dark.

Dark, just read their little explanation.

Let me see if I can get it

half accurate at least.

Dark matter suggests that it's out there,

but it's absorbing light or something,

it's not reflecting.

And it says that's actually

not what we're talking about here.

We're talking about something that is even,

well, it's really difficult to detect.

It's like it's a dragon in the garage.

Was it Carl Sagan?

No, it was Richard Dawkins, I think.

Carl Sagan talked about this

invisible dragon in the garage.

And now you can see part of

the problem here is in science.

There's a problem with with

pattern fitting.

That is,

you know that you observe something.

So what you do is you come

up with a theory that

explains that thing post

hoc after you've observed it.

Then, of course,

you go out and see if you can detect it.

and all our efforts to

detect it so far have failed.

Now I think the latest

effort is the best yet at

possibly doing this,

and if this one fails,

then we're going to have a problem.

Now the other option is to

rethink our theory of gravity,

that somehow it's more

complicated than what we anticipated,

that we're basically seeing

most of the matter,

but yet the gravitational

aspect of all that matter

might be different for some

reason than what we anticipated,

on a large scale maybe,

scale of galaxies or something,

the way the gravity works.

Don't ask me to figure out how that works.

I just laughed here.

I looked at Deep Burrito here.

Great question, Deep Burrito.

More insight and power than God.

Okay, well.

And then he follows it up with,

have we detected God?

Answer, no.

Well, I would say that we detect God.

Now,

this is just off the top of my head here,

so I'm basically pulling an

answer to my back pocket here,

Deep Burrito.

But I would say we detect

God in the same way we

detect dark matter.

We see the effects of dark matter,

and in this case,

we see the effects of the

existence of some mind

behind the universe, and that's

All of that is a huge topic

in and of itself.

So we would do it in the way

that forensic science works,

where forensic science

works with the effects.

Here's the effects.

What could possibly have done this?

Are there fingerprints here?

For example, we could ask that.

Are there any fingerprints

of God in the universe?

Why don't I ask ChatGP?

I asked Brock,

where does dark matter come from?

That'll make it think there

for a little bit.

Where does dark matter come from?

Yeah,

so it's telling me one Big Bang theory.

There are some leading ideas

of where it might have come from.

Big Bang theory,

so thermal relics from that.

Cosmological inflation.

And then particle physics

beyond the standard model.

This is way outside my

ability to comprehend.

Non-standard models.

Modified gravity theories, dark sector.

There could be an entire

dark sector of particles and forces.

Okay.

And cosmic string networks.

So hypothetical

one-dimensional topological

defects left over from the

early universe.

It could have fragmented

into dark matter particles.

But if it's eighty-five

percent of the universe,

how does that work?

Yeah, well, that's a big deal.

We could just say in layman's language,

this is a big deal about dark matter.

And I guess you saw its answer to,

are there any fingerprints

of God in the universe?

And ChatGP gives me a balanced answer.

So let's go into those

questions that it says under no.

So science can explain these

phenomena without invoking God,

though some questions remain.

Yeah, well, okay,

I think if you go back to,

they may actually, examples of fine-tuning,

counter-argument.

Yeah,

so counter-argument would be the

counter to the fine-tuning argument.

So this is where they would

say science can explain

that by invoking a multiverse.

And that has... We've talked about that.

We've talked about that,

and science is pretty

much... It backs it up.

Well, it's been debunked.

Where's the multiverse coming from?

Yeah,

that's the problem and I have an

article on my website I

should mention that I talk

about that and in that

article I have some references.

So it's one thing to watch

two guys here on livestream

say it's debunked or it's

been regarded as not

science or even a threat to science.

So I really do need to

provide references for that and

that you can find some of

those references in my article.

Oh, that article B,

something about the multiverse.

No, science and fantasy.

Look for my article that

talks about the fantasy and science,

and that'll have the references.

So I asked Grok, can nature create itself?

Oh, yeah.

And so it says, well,

it can create itself

touches on deep philosophical, scientific,

and cosmological issues.

Here are some perspectives.

One Big Bang theory.

Universe beginning with a

singularity approximately

fourteen billion years ago.

From this all matter, energy, space,

and time emerged.

However,

this doesn't fully answer how the

singularity itself came to be.

uh so it could have been

quantum fluctuations in a

pre-existing void but yeah

what created the void yeah

it's still really that's

chat points out

pre-existing multiverse and

then loop quantum cosmology

but again it doesn't really

it's just backing up the

problem one step right

It does back it up.

Now, the cyclic problem,

that loop quantum cosmology,

where is that here?

Yeah, it's...

suggests a cyclic model

where the universe

undergoes cycles of

contraction and expansion,

potentially self-sustaining

but necessarily creating

itself from nothing.

So I would like to ask a question there.

If we're at some point in

some cycle whatever, cycle X,

have we actually, in reality,

traversed through an

infinite number of previous cycles?

I already know the answer to that,

but from math, but, um, okay.

So if we're at a cycle in loop quantum,

so if we were at cycle X,

we're at cycle X,

we were at cycle X in the

quantum cosmology.

Does that mean, what's the question here?

Does that, we have actually, um,

Or, sorry, you could put, say,

physical reality or something like that.

We could mean physical

reality has actually

proceeded through an

infinite number of cycles in the past.

And colon, semicolon,

and then say emphasis on

the word actually.

Because mathematical models, no problem.

Has proceeded through an

infinite number of cycles in the past.

And we want emphasis on the word actually.

There?

I guess.

Let's see what it does with that.

Yeah,

it gives a little seminar on what

that is.

Finite number of cycles.

No beginning at WISEC here.

If we consider CycleX...

um finite number of sales

possibly first cycle

initial state yeah um

however how this first

cycle started you've just

pushed the thing back but

what about infinite cycles um

with no beginning, an infinite regress.

This has been a physical

reality that has indeed

proceeded through an

infinite amount of cycles in the past.

However,

this face is philosophical and physical.

Well,

I would say the biggest challenge is

mathematical because if you

have to count down through

a countable actual infinite

number of things and you

have to do it one at a time,

you can't do it because

there's a property of an

actual countable infinite,

which is what we would have

in the past if we've actually,

if you want to hypothesize

actual infinite number of

cycles as opposed to just

treating that whole past as

a mathematical object.

That's not the way things

work in physical reality.

So if you have a countable number, if you

The cardinality of a

countable infinite set does

not change as you remove

one cycle at a time from it.

It stays at,

and the cardinality is usually

represented by a symbol

called aleph subscript null.

So you never, ever get to X, cycle X.

no matter how,

there's no starting point either,

which is a bit of a problem,

but you never get to cycle

X. Mathematically, yeah,

you can make models that work,

but the problem is is

transferring them to

reality where you can't

treat the past as a mathematical object.

Instead,

you have to literally deal with

this problem of one cycle

following another cycle.

And let's say if each cycle lasts,

I don't know,

ten to the one hundred and

twenty billion years,

even if each cycle only lasted one second,

you would never make any

progress in getting to this

cycle right here.

But I think we probably lose

a lot of people.

I find whenever I start

talking about countable infinite sets,

cardinalities,

and traversing the actual infinite,

people lapse into unconsciousness.

yes um so I I just asked

like is infinite cycles

within quantum cosmology

possible using existing

existing mathematical laws

and it does say it they do

not rule out the

possibility of infinite

cycles but there are

significant theoretical

challenges and unresolved

questions regarding entropy

yeah exactly so like I said

brock's been pretty good

Yeah, it is pretty good.

I'm impressed.

And so this is where the

clarification I made is important here.

It is possible

mathematically to construct

a mathematical model where

you have an infinite number

of cycles and the

arbitrarity point X. Like

we work with infinite cycles.

Infinity all the time,

let's say when we're

integrating from minus

infinity to five or minus

infinity to plus infinity,

we do it all the time.

But it did point out here

that there are problems and

the problem one of them mentions is time.

How does time work and how

much time do you require at

each discrete interval?

That's the problem.

Then you start having to

work through that infinite

number one at a time.

So, mathematically,

no issue at all postulating

a universe that's

infinitely old in terms of

years or in terms of cycles,

mathematically.

But in reality,

then you have a major problem.

I think they say significant

theoretical work is needed.

That's baffle gab for...

we don't know it's probably

not going to happen in fact

yeah baffle gap by the way

is when you use technical

language speculative

extensions or

reinterpretations yeah of

these laws so what

qualifies as science and

what qualifies as science

fiction you got to make

that distinction here so science

You should be working with

things that are predictable

or at least reproducible.

You should be able to do some experiment,

testable, falsifiable.

But if you can't do any of that,

which you can't do here,

and you have a major

mathematical problem when

you transfer the

mathematical model into

reality where you have to

traverse each cycle one at

a time or whatever,

that's when your theory

moves into science fiction.

And you can write science

fiction using human languages,

but you can also write

science fiction using mathematics.

And the mistake some people

make is assuming that this

mathematical model,

because we said the word

mathematical and we used math in it,

that therefore this is

science scientific?

No, it's science fiction.

It's describing something

mathematically that in

physical reality is not possible.

Of course,

it does dodge and weave around

here with saying we need to

reinterpret and whatever

significant theoretical problems.

You can throw in words like that.

Well, it's hedging, right?

It's, okay, well, we don't have answers,

but we'll probably find them someday.

And I've got to hand it to

the AI thing here.

I've got to hand it to you

because it's learning how to, you know,

use baffle gab.

It's using how to dodge and

weave and work its way around the bush.

But in answer to Deep

Bruder's initial question,

have we detected God?

Well, I think chat GPT,

that answer we showed there, showed that,

well,

if we're just looking for the

fingerprints of God,

they're all over the place.

And then the argument then

heads over to the individual,

what are these fingerprints?

Because, I will say this emphatically,

I don't claim to prove anything.

I can, you know,

make mathematical proofs

where you're just working like, say,

with something as logically necessary.

But in real life,

we just talk about probabilities.

Is this more likely than that?

What's your rational

justification for this versus that?

And so sometimes you can't

even put numbers when

you're figuring out probabilities.

So Mama K is asking that same question,

is the probability higher

for the existence of God or against?

Yeah, ask that.

Did you ask that?

I haven't asked that yet.

I wanted to point out this,

because I asked the question,

where did nature come from?

And it basically gave the

same things that's always

been saying here, but in conclusion,

while science provides a narrative

for how our universe might

have come into being the

ultimate origin of nature,

whether it's our universe

or all of existence,

remains a profound mystery.

Profound mystery.

You know what I look for in

science when I'm reading a paper?

I look for what I call lack of data words.

You can put lack of data in scare quotes.

That's when you don't know something,

so you use words like might have,

possibly, narrative, profound mystery.

The best one of all that

takes the cake was one I

saw in a paper talking

about the origin of life.

They had

They had more than two dozen

lack of data words in there,

but the best one of all was, what was it?

It's logically possible.

Logically possible?

Wow,

like that really opens up the scope to

creative speculation.

There's a ton of things that

are logically possible.

It's logically possible that

a bucket full of gold

nuggets could fall from the

sky through the roof of my

house and land on the floor behind me.

That's logically possible.

It just takes a Boeing plane

full of gold buckets.

because those doors just

open on their own.

So you never know.

Yeah, but you never know.

My truck actually got hit

once on a trip by something

falling out of the sky.

There was nothing above me, no planes.

I stopped and I carved a

furrow in the side of my

truck as it ricocheted.

And you can see from the

angle of the furrow,

it had come from about a,

not a ninety degree straight down,

but an angled boat like this.

And it was about this big.

Could have been a little meteorite,

but it literally didn't

just chip the paint.

It literally left grooves.

It carved a furrow and left

grooves in the sheet metal of my truck.

So you never know.

Taking a walk outside could

be bad for your health.

It's logically possible you

could get nailed by a

meteorite and be off to the Purdy Gates.

So what about that question?

I tried to copy-paste.

How do we... It just disappears.

I'm asking,

is the probability that nature

began through intelligent

design or the Big Bang?

Or is the probability higher?

Oh, I guess it's already comparing.

It already knows what I was thinking.

It didn't even need my question.

Yeah.

A wealth of observational evidence,

including...

Well, you know,

it's pretty hard to talk

about the origin of nature

without having something

you could describe as a big bang.

Like, I'm not sure.

Let's say if God said,

let there be a universe.

Boom!

What would that look like in

retrospect as you look backwards?

I mean, you would see,

you could well see exactly

the same thing we're seeing today.

An expanding universe,

red shifting of galaxies.

whatever proportion of light

elements like it doesn't

really what you need is a

prediction that falsifies

one hypothesis while at

least verifying or allowing

that other hypothesis to

see the light of another

day so to speak so I might

ask um so it's interesting

though it's it's saying

there's the fine-tuning arguments there

Then it goes into probability.

Calculating a probability

for intelligence design is

not straightforward since

it's not based on empirical

science in the same way.

ID does not provide testable predictions.

Well, that's false.

It often relies on

philosophical

interpretations of complexity, order,

and purpose in nature.

It's subjective rather than

empirical measurement.

When you compare intelligent design,

the odds and probabilities,

with Big Bang or with the

guessing that's happening right now,

the theories,

what are the evidences that

they are saying that Big

Bang is coming at that are

actually testable?

Well,

one is the cosmic background radiation,

which they appeal to,

but I've read some other

papers more recently that

explain that cosmic background radiation,

and these are peer-reviewed

papers out of some journal.

I bet ChatGPT or Grok will supply them.

There's another one, though,

and that's the inflationary

period of the Big Bang.

The problem is,

is the more or less uniform

distribution of matter to

keep it from clumping early

on in the universe and then

it just collapsed so there

should be gravitational

waves produced during these

inflation during the

inflationary phase and thus

far we can detect we've

built very sophisticated

gravity wave detectors and

we can actually detect

colliding neutron stars for

example or black holes

as they send those

gravitational ripples out

through space so they're

that sensitive but we still

have not been able to

detect the massive kind of

gravity waves from the

inflationary period so some

scientists are saying well

okay that was a prediction

that falsifies at least the

inflationary theory in the

big bang but without the

inflationary theory we

We have a huge problem

because we shouldn't have a

universe like this at all.

It'll just collapse very quickly.

The matter will clump too fast.

You won't get this nice

distribution of stars and galaxies.

You know, on a related note,

before I forget, it said,

I wanted to argue with your

grok over there because it

said there's no, how was it?

It's not, ID is not testable.

No, it makes no predictions.

I think that was what it was.

Well,

here it says it's not within the

realm of scientific probability.

okay so here's a question I

asked chat gpt I said um

does what it says does the

design of a laptop computer

require intelligent design

and it said yes the design

etc etc and then it gave me

four reasons and the

interesting thing is the

first one is highly integrated systems

We see exactly that in living organisms.

It's like molecular machines

that have to be in place

before that thing can even

function as a living organism.

So it can't necessarily...

It has to co-evolve the

need and the supply at the same time.

We see so many highly integrated systems.

I read one paper in Nature

that says we may never fully...

Let's use the word reverse

engineer how the cell works.

It's just phenomenal.

It makes a laptop computer

look like something a

chimpanzee built in three minutes.

The second reason it says

that laptop computers

require intelligent design

is functional

interdependence where the

parts must be precisely arranged to work.

Again, in biological life,

this is the norm.

Without that, it wouldn't work,

without this functional interdependence.

Third is a clear purpose,

and they suggest processing information,

running applications, communications,

etc.

And there again,

let's take DNA for example.

We have the information that

You need to produce the

different kinds of proteins

to build the molecular machines.

All that information is

encoded in DNA for a purpose.

It's not just sitting there,

random bits of information.

It's got a clear purpose, and you say,

well, where did the machine come from?

All the different proteins

that we needed to assemble this, not we,

but the cell needed,

you have encoded information again,

such as computer code and circuit design.

Another paper in Nature said, basically,

we have molecular computers

built into the cell that

turn genes off and on

depending on the kind of

inputs they're getting.

And also, Craig Venter,

who decoded the human genome,

said it's time for biology

to start talking about the

information in DNA as computer code.

We have... So,

when you look at its four...

reasons or for identifiers

or for fingerprints or

markers for intelligent

design for a laptop,

all of those are present in

biological life and to a

much greater degree than

what we see in something so

barbaric and crude, relatively speaking,

as a laptop computer.

So simple and crude compared to the cell.

We have thousands of

scientists all over the

world working on gene regulation,

figuring out how the cell works,

molecular computers,

how these things get built

and are there at the right time.

may we're making progress

but the question is uh what

the percentage uh that we

know compared to what there

is to know and that's the

big question here's a good

one uh scum of the earth um

I don't know is that

aspirational that that name

or is it uh you've already reached

you know I kind of like

those names I used to play

uh back in my online gaming

days what was it halo and

my name was the clown I

called myself the clown

because I like people to

underestimate me yeah yeah

well that's that that's not

as bad as the clown but

scum of the earth actually

asks a very good question

very good well I think it's

to hold the earth down

Well, okay,

but there is actually a

scientific way to test to see if rocks

qualify or if they have the

marker of requiring

intelligence to produce.

So it goes through something

called functional information.

So if you said,

give me a peer-reviewed

paper discussing FI,

functional information,

it'll probably provide you a few.

I have worked with some of those papers.

One is in the Proceedings of

the National Academy of Sciences.

I've published a couple of

papers using functional information.

But the beauty of that

is that at the heart of the

function information is a

ratio of functional to

total possible configurations.

So the smaller that ratio is,

like the fewer,

let's say you want to break

into a bank vault and

nobody tells you the combination,

well you have a lot of

possibilities there.

and there's only one

combination will actually open it.

Now, if it's a cheap one,

you might have some slop in the numbers.

You might have nine or

twenty-seven different

combinations all in a very

narrow window that'll open.

So you put twenty-seven over

the total number of possible ones,

and you plug that into the

equation for functional information.

And I have that here,

because I just asked that

question earlier.

And it'll give you

number and that's how much

information is required to

to solve that problem of

breaking into the safe now

when you look at a rock you

say well what is its

function I look at rocks

out there and basically any

shape will do like there so

the ratio is actually very

close to one and

And when you have the natural law,

the actual equation for

functional information is

very close to one.

When you plug that in,

it requires zero

information to produce a random rock.

There's no laws of nature dictating this.

There's nothing.

It's just any shape will do.

So is a rock intelligent

design for a purpose?

A priori, you just go out there, no.

They're not because the

amount of information

required to produce any

random rock is zero.

And one of the markers for

intelligence is that it's

capable of producing huge

spikes in information or

effects that require a huge

amount of information.

Let's take the DNA, the genomes of life,

for example.

So once the amount of

information is higher than

a cutoff threshold,

and rocks fell well below that,

you can say, no, no intelligence required,

any shape will do,

we can just let stuff

happen and you'll get rocks

of any random shape.

I have a very practical

reason for rocks' existence.

I have children,

and when they were younger,

uh pockets became rock

holders yeah and they got

great joy from carrying

rocks yeah pretty when I go

to a beach and there's a

body of water and there's a

rock what do you do you

skip it yeah it doesn't

matter what the shape is

although you are trying to

go for the perfect shape

but it's how many times can

you make it bounce god loves fun

Yeah, but I would even,

like I've thought about all

of this stuff.

I've thought about it for years.

And what about these nice, beautiful,

round, but flat,

they want the pancake style of rock.

A nice one about like this, quite flat,

but rounded edges so it

doesn't dig in the water.

And what's the probability?

Because that's basically

that ratio in there.

that you'll get this,

and it might be depending

on what sort of beach you're looking at.

There's some beaches that

the physical system simply

will not produce those

kinds of rocks because the rock type,

let's say, is too soft or whatever,

it's just crumbling away.

But when you crunch the numbers,

it actually turns out that

nature is still capable,

just natural processes,

of producing large numbers

or enough of these skipping rocks

that very little information is required,

almost none.

The physical system will do that.

We're going on this rabbit trail,

let's come to the earth.

So a fascinating idea.

How do you measure the

amount of information in an

object or form?

Yeah, well, I actually did,

let me scroll up for you.

No, it's way, way,

way out of my sports ministry degree.

Okay, here it is.

If you show my window here.

Yep, it's on.

There it is.

So this is the functional

information approach.

to measure the amount of

information an object can form.

You have to know that it at

least has a function.

You don't have to know what

the function is.

For a lot of proteins,

we still don't know what

they actually do.

But when we look at the genomes of life,

we see that there are some

universal genes, for example.

We know that must be important.

We don't know what it does,

but it's got to be important.

It has a function.

So the next question is,

if you look at this, we've got ME.

Well, let's look at N first.

N is the total number of possibilities.

Total number of

possibilities for an object or form.

Now,

a real simple one is when you're

working with digital information.

Like, for example,

a bank combination or the

digital information encoded DNA.

Those are very simple to

work with because you have

a certain number of symbols.

And once you know that number of symbols,

let's say for DNA, it's four.

And then, well, for proteins,

there's twenty symbols.

and let's say the average

protein is three hundred amino acids long,

so it's basically n equals

twenty raised to the three

hundredth power.

So that's a large number.

But then we want to know Me,

and Me is a little bit

harder to figure out.

There's a way to do it,

and I published a paper,

and you can see it on my website,

a peer-reviewed paper on

how to estimate the

functional complexity or

the functional formation of proteins.

So you look at Me,

and you take that number,

and then you divide it by n

and then you take the log

to the base two because you

want to get it in bits

minus and that gives you

the amount of information

in bits now if you're

working with something

trickier where it's not

digital let's say the face

is at mount rundle then you

have to figure you'll have

to look at that you might

want to look at symmetry

you might want you you

still have to have some

variables here that you're

going to have to come up with

uh some estimates for so

symmetry might be one the

smoothness of the

topography versus all the

other rocks the faces at

mount rushmore yeah okay

you said rundle I was

trying to figure out there

oh sorry there's a canadian

version it's about two feet

high and that's just

trudeau oh yeah okay so

there are ways to go and

crutch in there it's not a

very popular place

There are ways to actually

estimate the functional

information required for

effects that are more complex,

like things that aren't

digital information.

You have to sit down,

you have to look at that effect,

and you could say, okay, so why,

let's say you landed at

Mount Rushmore from outer space,

you've never seen a human in your life,

no tourists around, you look at that,

and you say to your fellow alien,

that's odd.

And then, well, what's odd about it?

And so because you're scientists,

hopefully,

if you're flying interstellar spacecraft,

you say, well, here's something odd.

Look at the symmetry on

those four examples of what

we're looking at.

So you've got to come up

with what are the variables.

Then look at how smooth the...

sections are you don't call

them faces because you

don't see no human look how

smooth the sections are

versus everything else

we're seeing around us here

so gradually you come up

with maybe three or four

variables and you come up

with a probability of

seeing that kind of

symmetry in that area so

that would be the value for

one variable so that

equation you're seeing

there is actually a highly

simplified version of

shannon so do I have

shannon's more complex one

just above here um

No, they don't.

So people are afraid of equations.

So it's actually a sum where

you could take each

variable has its own sum and whatever.

And that's what I do for proteins.

And in the end,

then you get an estimate as

to how much information

that required to build,

even if you're a space alien,

never seen a human and that

you're a Rushmore.

Once you have that estimate,

then you can turn to your

fellow alien and you can say,

does this cross the threshold

where intelligence above

that threshold intelligence

is required and that

threshold is always a

number in terms of bits so

how many bits do we

estimate this required does

the physical system capable

is a physical system that

we're observing here

capable of producing that

level of information that

number of bits and it's

it's there's a way to do it

and then you come up with a number

I ran Mount Rushmore through Grok.

Could Mount Rushmore have

been created without intelligent design?

Because it previously told

me that a laptop could not

be created from a single

cell if given enough time.

Yeah.

Because it's not biological.

Okay,

so... And it was pretty clear that it

thinks that

Design is,

laptops and intelligence design

is needed.

But again, going back,

it was very clear that you

don't need intelligent

design for humanity or nature.

So... Well,

Mount Rushmore face is quite a

bit simpler than a human.

So what did it say about Mount Rushmore?

So this is interesting.

He said,

as it currently exists with the

card faces of four U.S.

presidents could not have

been created without intelligent design.

Here's why.

Artistic design.

Scale and precision.

Let's talk about the eyeball.

Geological and physical considerations.

So it's saying because it

couldn't have been a natural, well,

why not?

Material.

It's hard, durable rock.

Well,

that's really interesting because the

faces in Mount Rushmore

If you actually crunch the

information numbers,

a number of years ago, just for fun,

I think I came up with four variables.

You get a significant amount

of information required to

produce something like Mount Rushmore.

You need even more if you're

trying to make it look like

a particular person,

but let's say you're not.

You just want four human faces.

And it's saying here that it

does need intelligent design,

and it's giving the reasons,

but all of those reasons apply,

and even more so when it

comes to a human.

See, I have a face.

See, like that?

If my face was out in Mount Rushmore,

I think you'd probably say

you need intelligent design

to produce that,

as long as it's rock on a mountain.

But if it's a thing that has

billions of cells with

millions of molecular

machines all working

simultaneously to do this and that, no,

that doesn't require intelligent design.

Just read that conclusion.

Read the conclusion.

Unequivocal example of intelligent design.

Wow.

Yeah, well,

it's just too bad it doesn't

connect the dots.

I'm gonna ask it.

Yeah, why does a human face not required?

That'd be funny.

intent.

Oh, so there was,

it's kind of circular

reasoning here if it's

assuming there was no

intent behind the design

required to produce biological life.

Yeah, it's unlikely.

So it talks about

unlikelihood of natural formation.

What's the likelihood of the

Earth being exactly what's

needed to sustain life?

Well, you know,

I actually asked it a

question earlier on just

for one protein coding gene.

Just for one.

Like we have...

you know close to twenty

thousand protein-coding

genes plus or minus a human

does the simplest life form

probably has I I've seen

different papers with

different numbers I think

what was them two hundred

and eighty or something

like that different

protein-coding gene so I

said well what's the

probability how much

information actually with

that was the question how

much information would I

need for an average protein

and it did the crunch the

numbers let me see here

Final probability calculation.

That was a lower bound.

That was the simplest

protein you could possibly think of.

There it is.

Probability is ten to the minus twenty.

Extremely low,

meaning we would need

approximately ten to the

nineteen random trials.

Now,

I think reasonable is carrying a lot

of weight there.

Yeah,

and it's vastly underestimating the

information here because, well, it's,

I have,

I published a paper in Journal of

Theoretical, what is it,

Theoretical Biology and Medical Modeling.

where I actually crunch some

numbers for the average

protein and above and

higher and lower and it's

really underestimating the

amount of information and I

use real data.

I work with arrays that

might have let's say as

many as twenty thousand

different sequences for a

given protein family

So that's how many rows there are,

and then maybe a thousand columns.

So that's what I work with,

because that gives you the real data,

and from that you're able

to get a better estimate.

And it's usually around, let's say,

two hundred and fifty bits

of information.

Two hundred and fifty bits,

not sixty-four.

So I said, okay,

it said the higher FV

values was seventy-seven,

so it did it again.

So it's saying here,

each trial required one year, because,

okay, you can get evolution happening,

trial a lot of stuff if you

just work with bacteria,

but once you're starting to

work with life forms that

basically reproduce every

spring or once a year,

things really slow down,

and that's where most of

the diversity of higher life forms are.

So let's say,

and you just want to get one

novel protein.

Once you get, let's say,

once you get, say, a fish,

and that's all you have in the world,

just a fish,

but you're going to need a

ton of new novel proteins,

but it says just one, and just,

let's give that really

generous estimate that

ChatGPT calculated at seven bits.

If each trial took one year,

it says it would take that many years.

That's a lot of years.

That's more than a billion, I think, even.

Well,

the age of the universe is this many

years.

Therefore, it's saying...

effectively impossible

within any realistic cosmic time frame.

That's for just one protein coding gene.

Just one.

To get a life you need...

It's telling me that

There's no way that human

DNA needs an intelligence designer here.

There's no external design.

The complexity of the DNA

and its ability to code for

life is the result.

Keep in mind, though,

that a lot of these AI things,

they look online and they

repeat what they see.

Oh, yeah, totally.

So, I mean,

this is the reigning paradigm

within evolutionary biology,

and you'll see tons of papers.

Now,

you always have to look for lack of

data words in them.

The basic operations of

natural selection stuff, that's true.

You can observe that all the time.

But all that does is give

you variation within a

particular life form.

But to get, let's say,

go from an amphibian to a fish to,

let's say,

a figure skater at the Olympic Games,

you need a lot more protein-coding genes,

totally new novel ones.

Well, okay, that's a good question.

We're past our time.

Oh, we've got to quit, yeah.

So quick answer, no.

Let's answer this one

because this is a good one.

Yeah.

A quick answer is no as long

as the information is low enough.

That's why I talk about a threshold.

So if you take a bunch of

alphabet characters in a box,

dump them on the floor,

you'll probably get the

word ah because it's one letter.

So that requires no

information whatsoever.

But you could get a

two-letter word accidentally.

Right on the right side up.

Well, I mean, you're asking for a lot here,

especially if you start

getting to three- and four-letter words.

But whatever the physical system is,

you have to determine what

is the threshold that we

could naturally expect

maybe within reason to get that function,

whatever it is.

Let's say a two-letter word,

a three-letter word, a four.

Eventually,

dropping boxes of alphabet

characters on the floor,

there is a cutoff.

I would say, well, I don't want to make,

I just want to throw out

numbers out there.

Let's say fifty bits.

Fifty bits is pretty low,

but you will never in the

history of the universe

with people dumping stuff

get a fifty bit word.

out of doing that.

So it's, for every problem,

you look at the physical system,

what is the cutoff

threshold where randomly

and accidentally you can

produce information and how

much would that be?

So you get a probability number there,

you throw that into that

equation that you saw,

and that'll give you a

number and that becomes

your cutoff threshold.

And then you can double that if you want.

No, not double it, because bits are,

they go up logarithmically,

so you don't want to double it.

That's a huge increase,

but you can add a few more bits.

So if it's so little

information that you could

accidentally get it, no,

you don't need an intelligent mind.

That is why a threshold,

a cutoff threshold is required,

and biological life is

massively beyond that threshold,

far more than a laptop computer,

which Chad GBT concedes is

definitely an artifact of

intelligent design.

All right, well, this was a fun one.

We will see you next week, I think.

I think we're on for next week.

And yeah,

if you have ideas on what you

would like us to talk about,

any questions, throw them in the chat,

throw them in the comments.

And don't forget to hit the like button,

subscribe and hit the bell.

So that will not only notify

you when we have new videos coming out,

but it will help us

get our stream out there on

more people's algorithms so

they can enjoy math and

science and philosophy.

And I'm going to go and take

some aspirin because my

brain is hurting now.

All right.

Thank you, Kirk, for doing this.

Thank you,

Grok and ChatGPT for hanging out.

It does appreciate thanks.

I asked it.

Oh, yeah, well, somebody coded that.

we ask it if it requires an

intelligent design?

Oh, we never asked that.

You can do it yourself.

Yeah, everybody can do it.

Laugh at it.

Okay, talk to you later.

Okay, bye for now.

If I can figure out how to

get this thing working.

I'm punching that question in as we speak.

Well, maybe we'll just...

Yes, in a practical sense.

Yeah, okay.

It's hedging.

It's totally like,

I'm not going to let you

get away with this.

Yeah, I would then confront and say,

what do you mean practical sense?

All right, we'll talk to you later.

Yeah.