AI is a Big Fat Liar

Thoughts Off The Stem | Cannabis Infused Comedy

0:03

Welcome to Fox Office 10.

Thanks for joining the sesh with me, Justin Peroni.

I hope that life is being.

Good to you.

And this week I want to talk about.

The fact that AI is a big fat liar, Yeah, that’s right.

AI is lying to you.

And the longer that I’ve used it.

Because I’ve been using it for a few months now.

0:19

The longer that I’ve used it, I’ve realized it’s full of shit.

At first it.

Started off great because.

I was using.

It for SEO and what’s it called metadata to try and.

Move my podcast up.

The podcast ranks right, and at first.

It was great.

It was like Hussein Bolt racing white guys.

0:36

Right, but then after a bunch of prompts from ChatGPT.

All of a sudden it.

Fell off the Cliff like a wife who’d.

Been pushed by her husband.

To try and collect.

Insurance money.

Yeah.

Just right into the trees.

But the longer that I.

0:51

Used it the more.

I started to realize that it became the stereotypical yes man it was.

Like the the newest.

Stoner in the In the.

Circle that just sat.

In the corner and smiled and.

Nodded.

You know what I’m saying?

Yeah.

That’s what ChatGPT.

Is.

Or at least that’s what AI chat bots are.

1:08

They’re just, they’re just stereotypical, yes.

Men smiling and nodding.

We treat them as if they’re Gandalf.

The great, like they have some wizardly powers or something like that, but in reality what’s actually happening is we’re like.

New parents with.

Toddlers who are asking.

1:24

The toddlers how to parent.

Them, yeah.

So I initially started.

Using ChatGPT and I was using it like I.

Said to change.

The SEO, The metadata.

So that I could get a better reach.

In terms of.

Discovery and things like.

1:39

That right?

Well then after chat GT started doing what it does and justice agreeing with me, I started using Gemini and when I started using Gemini I.

Had this thought.

It’s.

Like when you meet someone cute and then all of a.

1:55

Sudden you meet their hotter sibling or friend.

And then you’re like.

Man, I wish I could have met you first.

That’s what it was like.

For Gemini.

For Gemini.

When I started.

Using it, it was.

Totally different.

It was like a It was.

A little.

Bit of a better set up I think.

And in using it, I found that the advice wasn’t as yes, Manny.

2:17

Now I started using Gemini to do the things that I’d done with ChatGPT.

To try and recover.

The failures of chat.

GPT and what I had.

What I what?

The rankings I had.

Lost from using that and when I started using Gemini I I now understood.

2:33

How to use?

Prompts.

Correctly and how to?

Talk to it correctly so that I could get a more devil’s advocate approach to it, right?

And so I started using that and I found it’s starting to dig me out.

Slowly but surely.

It’s digging me back out now the.

Problem.

2:48

With it is it’s.

Still got its flaws.

It still has its.

Issues because when I was using that.

Tool.

It’s still made.

Like simplistic mistakes, right?

And the fact that when you.

3:04

Ask it a.

Question.

We expect it to be like your buddy.

And doing doing the thing that you.

Asked it to do.

What ends up happening?

It’s a is it’s.

Actually, just like a, it’s like it’s it’s.

Designed.

To help to be helpful.

Right.

And in its design, it’s going to.

3:21

Basically lead.

You in a direction that you want to hit.

So it’s going to kind of.

Create a confirmation.

Bias to a certain extent, so you have to.

Learn what prompts to use and what.

Prompts not to use.

Right, because.

Unlike or I mean.

3:36

Just like Tesla.

When it first started and the car couldn’t recognize.

The end of a road the the ChatGPT.

Apps or like the AI chat app chat.

Bots don’t understand.

That they’re a tool to be used.

They’re they’re they.

They’re a robot.

And they should.

3:51

Just give you.

Some like unfiltered.

Unbiased opinions and play the devil’s advocate, but it doesn’t because it’s a technological.

Drunkard wandering the Internet trying to make friends.

So it’s.

Just going to.

More so tell you.

What you think, what it thinks you should.

4:08

You should have right?

So realistically, it just.

Becomes like.

A robot support animal.

That allows you to sort of like tickle it and pet it and get along.

With it and like.

Ease your tension.

And anxiety, that’s really what happens with it, OK.

4:25

So you have to kind of be careful.

With it and I think.

We all understand it at this point.

But we rely.

On it way too.

Heavily to give us.

Direction and insight on things like health.

Issues.

Or you know how to grow a.

Business now some of those things are good, some of the grow.

4:41

A business things are, but don’t trust it.

For health issues, we’ll find out why.

Later.

On in the dude for real.

Facts at the end.

Or the dude.

For real segment at.

The end of the.

Episode That’s right, there’s a few of.

AI gone wrong issues.

Or or scenarios that played out over time that we’re going to check.

4:57

Out.

OK, so we dismiss.

The fact that AI.

Still needs to be fact checked and and.

Spell checked and.

We.

Just we try to.

Use it because we’re.

Lazy.

And we want everything.

5:13

To.

Sort of just be easy for us, right?

So when AI was developed, we looked at AI as like our savior of.

Of how to fix things and how to do things and give us.

The right answers.

AI does not always give you the right answers.

No, no, you have to go in.

5:28

You have to.

You have to verify it for yourself, but.

A lot of people.

Just want to use it as a way to.

Confirm whatever.

Their thoughts and ideas.

Are because as.

Humans, like I said.

We’re lazy.

And we don’t really want to.

Put in that effort.

We just want something to sort of.

5:45

Solidify what we already.

Think and feel and and and how we want to add and like.

How we act is like the thing to do, right?

Does that make sense?

I feel like I went.

Off.

On a tangent, but.

As humans, we want to make.

Life super easy.

6:00

And we don’t.

Really think about.

Oh, double checking AI, It’s a robot, it should know what it’s doing.

So let me give.

You an.

Example of this OK.

When I was using.

When I was using ChatGPT, no sorry the example I want to give.

6:18

You is this one.

So I was using.

Gemini for the first time, but for a.

While right, so we had a.

Bit of a conversational rapport.

Going, I guess you could say.

And when I did, when I started having these conversations with Gemini, I knew how to.

6:34

Use the prompts right.

I knew how to use the.

Language right?

So I.

I formulated all my, my, I, my.

Questions and directions based on that.

But it’s still flawed.

Because I wrote, I got, I got Gemini to fix all the things that had happened with ChatGPT, right?

6:54

On a Friday.

I did.

This on I did.

This on a Friday.

So I.

Had ChatGPT or I mean I.

Had Gemini formulate.

How to get out of the hole?

That ChatGPT had made.

For me on a.

Friday set it all.

Up did all the things I was.

Supposed to do.

Changed all the stuff I needed to change within.

The podcast background and all that stuff.

7:11

And then went to bed.

And then on.

Saturday I woke up.

And I thought.

You know what, I better just double check.

This so I fed it all.

The exact same questions.

And it gave.

This was the response.

That it gave me.

It told me to return to turn all my stuff, all of the changes I’ve made to.

Revert them back to.

7:28

What I was doing?

Before.

So of course.

I was like.

What the fuck are we doing?

Here right so.

I I told it that I was like, what?

Are you talking about?

You just told me to make all of these changes and wait 2.

To four weeks.

So what do I do now?

And then it got.

Defensive.

Yeah, that’s right.

7:44

It got defensive on me.

And he goes oops.

Sorry.

I’m just a sentient.

Being trying to help you.

I understand that you’re trying to help me, but you’re not helping me in the way.

That.

I want you to.

Help me, you know.

I want you to be the devil’s.

Advocate in this.

So it goes.

I’m sorry.

Just leave all the changes that we did.

8:00

On Friday.

Don’t do anything.

And let it sit for a.

Bit and we’ll see what happens.

That’s what it told me and I was like, all right, fine.

So I’ve left it as it.

Is and I’ve seen an improvement to be honest.

I didn’t just jump the gun and change.

It so you have to constantly.

Go back and double check and make sure that what you’re.

8:17

Doing is the thing that you should be doing and that what ChatGPT or these AI chat bots are telling you.

Are the thing that you?

Should do, right?

So then another.

Conversation I have, I get it to build a picture.

And then it makes.

8:33

The pitcher and then we’re going through and we’re.

Changing things in the.

Pitcher, right?

Well, there’s a certain point where I.

Asked it a question about something that had responded with and it just started.

Creating the pitcher.

Over and over again.

So I said to it like.

Hey, stop making the.

Pitcher.

And answer my question.

And then it goes.

8:48

Hoops.

I’m sorry.

I guess I did.

The proverbial yes man move.

It’s kind of fun dealing with these things because they take on like a weird personality.

But yeah, so then it did that.

So you still have to check it, dude.

You have to absolutely check everything that you do.

9:04

With AI because most of the time.

It’s lying to you.

And I’m not saying that.

All AI is bad.

I’m just saying that.

We need to, really.

Make sure that whatever.

AI we use.

Or if we integrate.

It into like.

Making our lives easier.

9:21

It has a set.

Value.

It has set parameters to work under.

And then it can.

Only learn and.

Move in those.

Set parameters.

Like for example if you were.

To a company and you were going.

To put it in like a monitor or a sensor for your business, right?

Or or.

Some of the.

9:36

You know, research.

Modes that you have because.

Those things are beneficial to humans.

You need to have that you need to have.

That parameter, that requirement of you can’t move.

Outside this box, right?

If you let it run free, it’s just going to tell you.

Whatever you want.

9:53

So I do think that there is a place for AI.

In you know our.

As a human to make life.

Easier and better for us.

But we have to be careful about it.

And the best?

Places to use it?

Are in a business.

Where you’re using.

10:09

It for research and you have to research, you know, a whole bunch of data in in seconds or.

Or minutes or less time, or in a very short amount of time.

It’s good for sensors.

And it’s good.

For, you know, timers.

Things like that.

Anything that can that’s.

10:25

I.

Think is kind of is mechanical.

Now, I don’t think you should necessarily.

Have a robot that’s AI?

That’s a totally different story, though.

And the reason that I say this is because this week.

We’re bringing.

Back the weed facts.

Segment.

That’s right, we’re bringing back weed facts, baby.

10:40

And that is the.

Center of our weed.

Facts for this week.

Is how AI is improving the cannabis industry so.

Get ready because here.

Come your weed.

Facts for the day, baby.

10:57

Attention cannabis consumer.

So the weave.

Facts for this episode.

Are.

About AI and how it is going to help.

Progress the.

Cannabis industry so I.

Found this article from.

11:12

Cannabis risk manager dot.

Com.

And I’m going to.

Give you the highlights of the article because it was really.

Neat to and then in the IT was really neat to read.

Neat to read and see what they.

Were going to do with it and.

Then we’ll put.

I’ll put the link.

In the show notes.

So that you can check out the full article here, but there’s basically.

11:30

Four major.

Shifts and experts have.

Identified 4 optimizations.

That are being made and integrated by.

A or by integrating.

AI into the industry and it’s going.

To enhance operations.

As well as the.

Consumer experience and of.

Course, this is what.

11:46

They say, and of course.

It will come with the opportunity, it’ll come with.

Opportunities and some challenges.

So the the 1st.

Major shift comes with cultivating cultivation systems.

Generally, there are laborious.

And intensive experience driven.

12:04

Process you have to.

Balance light cycles.

Humidity, nutrients, and genetics to produce consistent.

Yields and quality.

Flour and any.

Producer could tell you that this takes up a lot of.

Time, and there’s a lot of.

Experience that goes into knowing how to do that.

So head growers.

And things can really benefit from.

12:20

These systems, especially if.

You’re new to the industry.

Just out of university or college?

And you can it’ll.

It basically takes away having to learn decades of of of tiny little adjustments to make your product perfect, right?

12:36

And then it can.

Set up predefined parameters.

For your grow up.

And notify your production.

Team at the exact.

Right moment.

To make those adjustments.

Where you would just.

Where normally.

You would need somebody that had so many years of.

Experience they could.

Look at it.

And be like, oh, I know what’s.

12:52

Happening.

So incorporating AI sensors.

Into grow rooms will.

Introduce the predictability.

And consistency right down to when the plants.

Are most vulnerable.

To pests and diseases, which is one of the.

Biggest.

Problems in terms of growing a brand new crop right and having.

13:08

To having to.

Make sure that the yield is right as well.

There’s also the predictive.

Analytics in retail.

That are going to.

Be affected by this.

So basically they use.

Predictive.

Analytics platforms already, but by that analyze purchasing patterns and product preferences and market trends.

13:27

OK, but as they become.

More.

Sophisticated and integrated into.

Point of sale systems.

The they become They’re going to help dispensaries anticipate, demand, optimize.

Inventory and then refine marketing campaigns.

13:42

More accurate, more accurately.

For producers.

So there’s some, like I said it within the box.

Of the parameters.

That you can just define it.

Or you can define.

For it, it’s it’s great at doing that.

But when you let it run Willy nilly.

You never know you.

Might just get like a little shop of horrors weed plant that just starts.

13:59

Ingesting everybody.

Maybe they try to.

Smoke us.

The benefit to the?

Consumer is that.

Recommendation is recommendations because.

AI will be able to suggest products.

Based on their user preference, their desired.

14:14

Effects and even medication, medical.

Conditions in some cases.

And it has potential.

To be placed into apps.

Or wearable.

Devices that people wear.

So that it starts.

To remember record.

And keep the data so that you can you can next.

Time you go into a pot shop.

14:31

You could just easily.

Look at the data, be like Oh yeah, that strain did.

This and that’s what’s happening to me right now.

I have pain in my knee and I need something that’s going to take away the.

Pain.

It’ll be great.

The downside for that will be that the.

People that.

Don’t want to be?

On a.

List and show.

Their ID Guess what, you’re on the biggest list ever, baby.

And then obviously in the.

14:49

Direction of in the.

In the.

Section of quality.

Compliance and control.

There’s going to be, there’s going to be a.

Lot more.

Ability to refine how quickly you can do.

The reporting of things.

15:05

So when you’re getting your shop ready, when you when you have to do your.

Annual, I think.

It’s annual, maybe.

Your reoccurring.

Compliance when you have.

To fill out that.

Paperwork AI can help.

You fill out that.

Paperwork.

It’ll also make sure.

That it’s correct.

15:21

And that.

Everything that you’ve got is what you need so that there’s no missed compliance issues.

There’s no it cuts down on fines and and.

It’ll it’ll really.

Bring down mistakes basically in terms of.

Quality control you.

15:37

Won’t run into.

The issue where people.

Are.

Lying about percentages, putting on bad like wrong labels and things like that.

Miss notifying.

People of what?

Is actually.

In the in the.

15:54

Package that they’re buying right.

So that’s.

Going to be.

Very good.

That’s going to be.

Very beneficial.

It’ll alleviate.

Discrepancies and reduce the.

Potential for non compliance.

16:10

And regulatory fines as well.

I mean, that’s basically what I just said, but that’s what that’s what they’re talking about.

It’ll help with research.

It’ll help.

Refining R and.

D by having AI.

Analyze extensive data clusters trials.

Studies and consumer feedback.

Producers will have a tool to identify new.

16:27

Opportunities quicker.

Yeah, because how long does it take you to go?

Through an article.

AI could do it in like milliseconds right?

It’ll help producers.

Refine and identify cannabinoid and terpene combinations for.

Specific therapeutic outcomes.

Which is clearly beneficial for the consumer.

16:44

I mean, obviously.

And then?

Ultimately AI in.

The R&D phase of production.

Has a multitude of applications.

To better serve.

The industry as a.

Whole.

Helping companies innovate, reduce.

Time reduce time wasted on trial and error and maintain.

Quality.

17:00

So again, going back to.

You know, trying to.

Like.

When you do like test grows and things like that to see if.

You can get a.

Phenotype to work.

That’ll be.

Perfect for that because it will help you.

It will help you in the.

Process of that and potentially cut down time as.

You learn.

So those are your refacs for this week.

17:18

How?

AI is going to help the cannabis industry.

Grow and.

Flourish as a whole.

That’s right, attention.

Cannabis consumer attention attention cannabis consumer Now The thing is with anything like.

This OK?

You’re going to run into issues.

17:33

Where you have to.

Research or you have.

To double check you have to triple.

Check sometimes.

What the AI does for you because.

There’s going to be.

A big learning curve.

Like when AI first.

Came out, everybody thought, Oh my God, we’re just going to.

Throw AI out there and it’s going to be amazing, right well, the.

17:52

AI had a few flaws.

Yeah, that’s right.

AI had a few, some really bad.

Ones and some really.

Funny ones, but it’s.

It’s because of the.

Integration with humans.

That it’s funny and dumb.

OK, because humans don’t.

Want to actually put effort into doing things?

18:09

They just kind of.

Do things because they’re told.

To a lot of the time.

Most people, I think.

Just negate.

The fact that they should put, they should put some time.

And effort into understanding what it.

Is that they’re doing and what they’re.

Using SO in the beginning of AI, these are some of the dumbest things that have.

18:24

Happened to people or that people have used it for or AI fails in general and this is the do for real segment.

Baby dude, for real, for real.

Dude so this is dumb things.

That people have done.

With AI now I’ve.

Got a few of these.

18:40

Instances or stories from.

Around the web.

The first one comes from vice dot.

Com again I will put these in the show notes.

As well.

So you.

Can see them so you can actually go and check out.

The different sites that have.

More than.

Just these ones, but I thought these ones.

18:56

Were kind of the.

Funniest so vice dot.

Com’s the 1st.

One and it’s.

An article called.

Stories for people who asked AI for.

Medical advice A. 60 year old man.

Wanted to reduce salt in his diet.

Chat ChatGPT told him to.

Replace table salt.

With sodium bromide.

19:13

Sodium bromide is a.

Chemical used to clean.

Swimming pools.

He did it.

And for three months.

He was hospitalized with.

Bromide.

Poisoning.

He suffered hallucinations and.

Confusion and his case.

Was documented.

In the Annals of.

Annals the annals of internal medicine clinical cases so he became.

19:34

One of the first.

People, that was.

Documented to have a stupid health.

Problem because of AI.

This one comes.

This one involves potential.

Murder suicide.

OK, so if you don’t, don’t listen to this.

But it was it was dumb, so I had to put it on here.

So at cio.com.

19:51

They had an article.

Called famous analytics.

And an AI and.

AI disasters, so this.

One ChatGPT, allegedly.

Involved in a murder?

Suicide in August of I think.

The year was actually. 20 It was 2025 in August 2020. 5 so.

20:08

Relatively new this.

Wasn’t way back then.

This was new, so according to.

The New York Post report.

ChatGPT.

Have, may have, may have.

Fueled the delusions of.

A former Yahoo manager who killed his mother.

And himself after months of interactions.

With the chat bot.

Whom he called Bobby.

20:25

Stein, Eric.

Solberg, 56, killed his mother, Suzanne Everson Adams. 83 in her home in Greenwich.

Connecticut on August 25.

August 5th, 2020.

Five and then.

Committed suicide shortly after Solberg, who developed illusions that his mother his mother was a Chinese intelligence.

20:43

Agent asset who attempted.

To poison him with.

Psychedelic.

Drugs through his car’s air vents shared.

These thoughts with Bobby for.

Months and the chatbot.

Allegedly.

Agreed with and confirmed Solberg delusions.

20:59

So basically this guy.

Sat around paranoid.

Out of his mind, probably with, you know, some issues that he needed.

To deal with.

And the chat bot told him, yeah, no, go ahead, kill, kill, kill.

Do it.

Just go right ahead.

This one.

Is from BuzzFeed dot.

21:15

Com 10 times AI and robotics have done horrible things.

So this was an AI in a robot.

In a factory, I think yeah, at an at.

An agricultural I can’t.

Speak right now.

At an agricultural produce facility in North.

Korea An.

21:31

Employee in his 40s was inspecting a.

Robot’s sensor operations.

When the machine suddenly.

Malfunctioned.

In a horrific error, the robot.

‘S arm grabbed the man, shoved him against a conveyor belt and crushed his face.

Decent chests he was.

Rushed to the hospital.

21:46

But died shortly after.

Officials believe the robot confused the man with the box of bell Peppers.

And it had been programmed that it.

That it had been.

Programmed.

To handle one report.

From the Korea.

Herald quoted A city official is saying the robot was responsible for lifting boxes of produce, it appears.

22:05

And misidentified the man as a box and grabbed him.

So, yeah, right, maybe.

You should have checked.

That code.

Before you put.

It in there?

Before you before you had robots moving heavy things around.

Or deciding that humans were.

A box of bell Peppers.

When I was saying.

22:23

Sensors.

Earlier I really just meant like little sensors.

That sit on things and.

Sense things, not a robot, OK?

So this one’s from reader.

Sdigest.com.

And it’s called hilarious.

Mistakes made by artificial intelligence.

So.

22:39

Hanson Robotics CEO David.

Hanson and his humanoid humanoid robot Sophia.

Appeared on the CNB.

On CNB CS The pulse.

He himself.

Asked the AI.

What was clearly.

On the.

Mind of many people in the studio.

22:54

Sophia, do you want to?

Destroy humans without hesitation, Sophia smilingly, A tad too broadly.

For our.

For these people’s tastes, responded.

OK, I’ll destroy humans.

Not so bueno.

23:12

Live Science.

Had an article called. 32 times artificial intelligence got it catastrophically.

Wrong a 20. 18 Analysis by the American Civil Liberties Union.

Oh, this is a.

Government one, this one.

‘S kind of.

Good A20. 18 analysts analysis.

23:29

From the American.

Civil liberties.

Union.

Found that Amazon’s Amazon’s recognition AIA.

Part of Amazon Web Services incorrectly identified 28.

Then members of Congress as people.

Who had been arrested?

23:45

The errors came with images of members of both main parties, affecting both men and women and people.

Of color were more.

Likely to be wrongly.

Identified great racist AI right off the bat, baby.

So that’s why we have to fact.

24:00

Check this shit.

You got to make sure that it’s right.

Before you put.

It out there, man.

So last one.

This one’s like an.

Honorable mention, OK.

So vice.com had another.

This was a second.

One I pulled from that same article.

Of. 4 horror stories from people who asked AI for medical.

24:17

Advice a Moroccan man.

Asked ChatGPT.

About a cauliflower.

Like Lesion near his.

Anus.

The bot mentioned hemorrhoids.

And suggested elastic ligation.

A.

Procedure that uses a rubber band to cut off blood.

24:34

Flow to swollen veins.

He attempted it himself with a thread.

What?

People are nuts dude.

Doctors later removed it.

After he.

Arrived at the hospital.

In agony.

As it turned out, the growth wasn’t a hemorrhoid, but a 3 centimeter.

24:51

Genital wart.

Dude.

For real.

Dude for real.

For real dude.

So if we’re.

Learning anything.

From this.

Don’t trust AI, don’t let it be like the.

25:08

End all, be all.

Of whatever it is that you’re.

Doing make sure that you put some effort in to learn.

About the thing that you’re trying to.

Accomplish and also.

If you have medical if you are asking it medical advice, go.

To a doctor AI.

Doesn’t know it’s a It’s a chat robot.

25:25

Don’t, don’t.

Rely on AI.

Those are my thoughts.

Off the stem for this week.

Hope you enjoyed the sesh.

Hope you come back next week with me, Justin Peroni on thoughts.

Off the stem.

You know what?

You can now.

Go to TOTS 4. 20.com and get everything right there that’s.

Right, so just go.

25:40

To tots420.com and check out all the episodes.

And all.

The links to Spotify.

YouTube, all that.

Fun stuff, it’s all.

Right on that page.

And it’s ready to.

Go.

You can also get.

The.

Show notes.

I’ll put those up.

As a blog post on there so you can read through.

25:56

What I just read.

You and then you can have all the.

Links to all the places I’ll also put.

The.

Links up in the weed facts section.

Oftots420.com and I’ll put the.

Dude, for real facts.

Up in the dude for real.

Facts section of.

Tots420.com.

If you made it this far.

And you?

Enjoyed the sesh hit?

Subscribe, Click, Like Leave.

26:12

Me a comment.

Tell me what?

You think and.

Until next time.

Keep your lids low, baby.