GPT‑5 se souvient de tout (et ça fait peur…)   #chatgpt #opeani #gpt5 - Vidéo Dailymotion

#ParlonsIA CHAT GPT - Perplexity -Claude

1 like de la video = 1 merci ❤️   MES FORMATIONS →   https://parlonsia.teachizy.fr/ Sam Altman nous a prévenus : ChatGPT est une bombe à retardement. Je vous montre le hack pour entrer dans sa mémoire et révéler tout ce que l'IA sait sur vous. Une vérité choquante sur les risques pour votre vie pro et perso.  #chatgpt #opeani #samaltman  #chatgpt 5 ##gpt   🔗 Rejoins la communauté IA & Business 🌐 https://parlonsia.teachizy.fr 📺 https://www.youtube.com/@IAExpliquee.x 📺 https://www.youtube.com/@ParlonIAhizy.fr 📘 Facebook : https://bit.ly/4kabhuA 🐦 Twitter / X : https://x.com/ParlonsIAx 📩 Contact : formation.ai87@gmail.com 🎙️Podcast – https://spoti.fi/4dqZ3uO ✍ BLog: https://medium.com/@flma1349/ 💃https://www.tiktok.com/@parlonsia 🎁 Gagner des Formations : https://bit.ly/cadeauxIA --------------------------- IA a tester :  Coder agent IA: https://bit.ly/Coder_agentiA Short AI: http://bit.ly/4lzE782 SEO agent IA : https://urlr.me/P8AS5N code rection 25% : PARLONSIA25  ⌚ Timeline Vidéo 0:00 : ChatGPT, une BOMBE à retardement  ? 1:10 : L'Avertissement de Sam Altman que personne n'écoute. 3:05 : Pourquoi vous avez PERDU 3 ans à mal utiliser l'IA. 5:27 : Le grand MENSONGE de l'IA "experte" 9:15 : Les influenceurs IA vous ont menti (et ça leur déplaît que je le dise). 11:40 : Le HACK : Je rentre dans la mémoire de ChatGPT (Démonstration). 14:02 : Ce que l'IA sait VRAIMENT de vous (c'est terrifiant). 16:25 : Patrons : Vous êtes assis sur une MINE D'OR  18:30 :Comment l'IA peut RÉVOLUTIONNER votre travail.   ---- ChatGPT, Sam Altman, sécurité ChatGPT, OpenAI, mémoire ChatGPT, hack ChatGPT, danger IA, preuve juridique, conversations ChatGPT, secret professionnel, données personnelles, mauvais usage IA, LLM, prompt engineering, AGI, Jarvis, intelligence artificielle générale, marketing IA, comment fonctionne l'IA, NLP, fine-tuning, apprentissage par renforcement, GPT, Generative Pre-trained Transformer, modèle Transformer, Vaswani, All you need is attention, prédiction de mots, compilateur statistique, influenceurs IA, diffamation, classification de données, profilage psychologique, profilage consommateur, quand c'est gratuit c'est toi le produit, formation IA, IA pour entreprise, IA pour salariés, IA pour patrons, risque politique, HPI, DSA, IA pour thérapeutes, IA et psychologie, IA et médecine, data analyst, productivité avec l'IA, optimiser son temps, hack memory, prompt secret

Transcript

00:00Hi there, today I'm going to show you how to hack ChadGPT's memory.

00:06This is a crucial topic. Sam Atman, CEO of OpenAI, director of TchadGPT,

00:13explains to us that the information collected with TchadGPT can be used by the courts against you.

00:21Did you realize that you were in the presence of the worst slut?

00:26with whom you share office hours and personal life?

00:31These are the topics we are going to cover in this video and we are going to talk about a key point.

00:36Haven't you been lied to about ChadGPT's capabilities and what it can do?

00:43Yes, yes, I'm talking about prompt engineering.

00:46Is simply asking ChadGPT “Open me the best sources with best practices for my professional field”

00:53wasn't the biggest lie in the AI hype?

00:57Isn't AI a kind of big speculative bubble?

01:01which actually turns into a trap,

01:05that of having bitten into a seemingly free product?

01:11Go fuck your life.

01:13Just imagine for a moment the AI that is today capable of connecting to the mailbox,

01:17to contact your employer and tell them

01:19"Your employee did not stay as many hours as he says on the file."

01:23or even “To your wife”.

01:25He exchanges a lot of emails with Martine.

01:28He talks about holidays and weekends.

01:30This is not professional content.

01:32Hey, it's starting to sting, but that's what the models are capable of doing.

01:36So, how can we avoid these types of problems?

01:38But as an employer, you also have every interest in knowing one thing.

01:41“How to use the data of your customers, your employees,

01:46to optimize the way of producing content?

01:50These are all the topics we will cover in the “Prompt Engineering” section.

01:53We're going to go much further than tutorials that simply offer you hype.

01:58This is where we make the difference.

02:00This is happening now.

02:00Sam Atman, CEO of OPDI, tells us this very clearly.

02:04Conversations are not protected by professional secrecy.

02:08So what not to say to ChatGPT?

02:11Don't confuse him with your lawyer, therapist, or doctor.

02:16There are things that as a ChatGPT user,

02:20You will discover what the models use your data for.

02:24What should you not say to make a model your worst nightmare tomorrow?

02:30You don't realize you're sitting on gold.

02:34And this chatbot became the best way for you to sell products.

02:40It is these methods, these tools that we are going to analyze, discuss.

02:44and show you how you need to interact to be able to access this information.

02:49So, welcome to the ChatGPT memory hack.

02:53What is AI about you?

02:55If it's free, then you are the product.

02:58We are about to descend into the memory of ChatGPT

03:02to understand what he knows about you and understand what the risks are for you.

03:07But as a company, wouldn't you also have an interest?

03:10to use this type of information for your customers?

03:13I'm going to talk to you about these topics and I think that to understand how the models work,

03:18We need to really understand what they are capable of.

03:20and to get out of what I call hype, or rather network marketing.

03:25Well, knowing that you are facing a machine that is your best and worst enemy,

03:32I know many of you are disappointed with the use of AI

03:35and say to me, "Yeah, but AI is a bullshit tool in which, in fact,

03:40You put in instructions and you actually get anything at the output.

03:43Well, actually, you're not wrong.

03:45And I'll tell you, it's not really the AI's fault,

03:48It's rather the fault of those who showed you how they use AI.

03:53So, I'm going to show you what I don't advise you to do.

03:56and the reason why you need to change your strategy.

04:00You just lost 3 years.

04:02How much money do you consider 3 years of life, 3 years, your incompetence is worth?

04:09Others today are able to know how to actually use it,

04:13but certainly not in this way.

04:15I will explain.

04:16Are we in the presence of Jarvis?

04:18You know, that super AI, in all the videos, you hear,

04:21we are in the presence of AGI, Artificial General Intelligence,

04:25who is able to make decisions, to do your job,

04:27that you're going to be fired, your boss will only use AI.

04:31And you, as the boss, you tell me,

04:32I'm going to lay off my employees, I'm going to take on AI,

04:34because they are able to answer mail, send email,

04:37do marketing, search the Internet, in short, they are capable of anything.

04:41But is it really true?

04:43You realize that if you no longer have employees, it is not possible.

04:46For what ?

04:46Because in fact, you are not shown how to use it.

04:49What we're telling you today is that you have an expert.

04:51Each time, you have to call them assessment experts, and they tell you.

04:54We take an example of these prompts which are in the best training courses on the web,

04:59which are seen by 200,000, 300,000 people.

05:02So it's considerable.

05:03These people have an impressive power of communication.

05:07on the networks, and they are working on it.

05:09They work on it so much that they try to make sure

05:12that people like me can never reveal to you what I am about to do.

05:18And yet, at some point, you have every interest in discovering the truth.

05:22The truth is, you can't ask AI to be both a content creation specialist,

05:28entrepreneurship, investment, creativity,

05:31and you ask him to have all the skills to properly evaluate all the subjects.

05:35All topics.

05:36You have Jarvis.

05:37You are touching the heartstrings of people who have no skills.

05:43and who think they are dealing with a machine or almost a person

05:47who can also provide relevant advice and constructive feedback,

05:51to note down ideas, experience.

05:54But I was going to ask you, what is the experience of an AI?

05:57How do you define an AI's expertise in a field?

06:01How does she go about defining the right strategy for a professional field?

06:06And this is where we get into how an AI works so that it responds to you.

06:11Does she really have the ability to go into entrepreneurship files,

06:15to search for the conditions, the rules, the comparisons, the methods, the capis,

06:20then go and make an efficiency comparison with other companies

06:24and give you constructive feedback?

06:26That, in less than 0.2 seconds?

06:29It's a bit complicated, isn't it?

06:30Don't think.

06:31Don't think that perhaps the point of disappointment of AI in the professional field

06:35rather does not come from the way you were sold a product

06:39or rather you were made to believe,

06:41which was not at all possible with these models.

06:43So what are these models capable of doing?

06:45before talking about TchatGPT's memory?

06:48To understand what these models are used for

06:51and why they were developed

06:53and how to use it,

06:55we must understand that first of all,

06:56the difficulty was getting out of it

06:59to understand human language.

07:01It's not an easy thing, it's extremely complicated.

07:03and it must be said that the French language

07:05is even more complicated compared to other languages.

07:08If we say the avocado eats the apple,

07:11we are on homophones,

07:12that is, the words have the same sound,

07:14avocado, the fruit, avocado, the profession.

07:16We don't know if it's an avocado, a fruit that eats a fruit

07:19or a person.

07:20So, we had to do what we call NLP,

07:23Natural Language Processing.

07:25From there, we had to develop technologies

07:27which, based on parallel processing models

07:31and different models,

07:32including the Transformer model,

07:342017 with Vasani,

07:37who invented what we call

07:38attention mechanisms.

07:40What we teach the model to do,

07:41it's in a sentence.

07:42What are the keywords?

07:43The super important words,

07:45we keep it simple,

07:46to understand what you are saying.

07:48All this will help us understand the word

07:49in relation to other words

07:50and to understand if it is the lawyer,

07:53of fruit or of a person.

07:55As a result, we will understand the sentence

07:58and we will understand the sentence in a paragraph.

08:00But you understand one thing that is super important,

08:03is that the computing power is limited

08:05and the number of words that defines the sentence

08:09or the texts you are going to send

08:11will make sure that each word is identified

08:14in relation to other words,

08:16of all other words.

08:17And so, at some point,

08:18there is an explosion of calculation.

08:21And yes, attention blocks and memory

08:24and processors are not infinite,

08:26mathematical methods had to be found.

08:28And so, we're going to get into the game

08:30of simplification of logits.

08:32So at some point,

08:33there are mathematical models

08:35which necessarily make simplifications.

08:38And that will lead to a few errors.

08:41So, we will have to be precise,

08:43much more precise than you are led to believe

08:45in the discussions you see with AIs.

08:48So, the models,

08:49they were pre-trained on a lot of data.

08:51That means they've read millions of texts.

08:53to say, that's math,

08:55that's physics,

08:56that's science,

08:57That's finance.

08:58And at some point,

08:59to make them very good,

09:00we do what we call fine tuning.

09:02But do we do fine tuning?

09:03on all models?

09:04Well no, it costs a lot.

09:06because you have to select the data.

09:08The data, essentially,

09:10It was stolen from the web.

09:11But on the web,

09:11there is everything and anything.

09:13And you have good examples of anything.

09:16From everywhere, the models will therefore feed

09:18of knowledge written by humans

09:20and today also written by AI.

09:22When you see that from everywhere,

09:24you have marked

09:24you are an expert in the art of doing something,

09:26well from everywhere,

09:27AIs will replicate this pattern.

09:29Is it that for all that,

09:30that's good, that's useful,

09:32it's something relevant

09:33to be expert, expert, expert?

09:35Well the answer is no.

09:37You will not see anywhere,

09:38in a prompt system,

09:39you are an expert.

09:40You are the best system in existence.

09:41You're Jarvis number one.

09:43It doesn't exist.

09:44And when we go into this type of prompt

09:46where you are told

09:47you will make money

09:49and you're going to give me a x10

09:51on my investment

09:52giving me all the advice,

09:54Well, unfortunately,

09:56you only have that,

09:57Today,

09:58which drives the models.

09:59So don't be surprised

10:00that the model gives you

10:01the same type of response.

10:03For what ?

10:04Because these models actually have

10:06a capacity

10:07to calculate probabilities

10:09to use words

10:11which are usually used

10:13in this type of context.

10:15It is a prediction model,

10:16but not only that.

10:17They were also trained,

10:19Today,

10:19to make decisions.

10:21Rather than decisions,

10:22we will say,

10:23that they are capable,

10:24through learning

10:25by reinforcement,

10:26A,

10:26when you don't know

10:27express yourself very well,

10:28two,

10:29know what the words are

10:30the most important

10:30in this sentence,

10:32which gives roughly

10:33suitable results

10:34in one, two, three conversations.

10:37As soon as it's bigger,

10:37you realize

10:38that it goes to hell.

10:40And then,

10:40it allows them to know

10:41that when they don't know

10:43how to do it

10:44or that you ask him

10:45an image or a video

10:46or music,

10:48well,

10:48they are capable today

10:49to go get tools

10:50and therefore,

10:51they have a decision-making process.

10:52Are these models,

10:53Today,

10:54are really capable

10:54to do what you ask them,

10:56In any case,

10:56in the manner

10:57which we plan

10:58the information?

10:59Is it if you ask him

11:00does content creation,

11:02of entrepreneurship,

11:02of investment,

11:03is the model

11:04really, really goes

11:05check the country's legislation,

11:07know what your business is,

11:08what are the best strategies,

11:10go and make a comparison,

11:11look at a forecast

11:12according to trends

11:14of the market,

11:15well,

11:15we really messed up

11:16out of your mouth.

11:17I tell you very clearly,

11:18No,

11:18they're not going to do it.

11:19And so,

11:19No,

11:20you don't have a general meeting,

11:21you don't have Jarvis.

11:22SO,

11:22every time you are told

11:23attention,

11:24the model will replace you,

11:26Good,

11:26I think that

11:27these pipo players,

11:28at one point,

11:29it would be necessary

11:29to understand

11:31that we are not at all

11:32on autonomous systems.

11:34We are on compilers

11:36of statistical words

11:37advances,

11:37Yes,

11:38they are advanced,

11:38but they are not able

11:40at all to do these tasks.

11:41They will just peck

11:42a few words inside the text

11:43to understand

11:44what is most important

11:45and throw you an answer.

11:46Is the answer true?

11:47No,

11:48It's fake from A to Z.

11:50Your AI model,

11:51it's your worst enemy.

11:53And yes,

11:54he learned

11:54to pretend

11:55to understand you

11:56and pretend

11:57to answer you

11:57he knows so well

11:59imitate human language.

12:00That's why

12:01that he was trained.

12:02He was trained

12:03to find the words

12:05the most provable

12:05in a context

12:06and therefore,

12:07you don't see anything.

12:09But the true nature,

12:10these are systems

12:11of classification.

12:13And there,

12:13we switch back

12:14on your profile

12:16from memory,

12:17cat memory.

12:18With the magic prompt,

12:20pardon,

12:20the hack prompt,

12:21I don't know what to call it

12:22so that it doesn't sound too bullshit,

12:23but in fact,

12:24you understood,

12:25we can use

12:26real skills

12:28to go inside,

12:30In fact,

12:30data

12:31which are of the system

12:33to extract

12:33what the model

12:34it's about you.

12:35SO,

12:36the model will tell me,

12:36I will give you an example,

12:38what level I am on.

12:39Today,

12:39I'm on a free plan,

12:40I don't pay for it,

12:41I use other models

12:41by paying,

12:43and several others besides.

12:44That he has my IP address,

12:46he knows my frequency of use,

12:48than the operating system,

12:49it's Windows 10,

12:50Chromium,

12:51that I use an average

12:53of message words,

12:54my tone,

12:54my goals,

12:55my tongue,

12:56and we are in the process

12:57to return,

12:57In fact,

12:57in my privacy.

12:59I am.

13:00You are in front

13:01of a machine

13:01who is capable

13:02to say roughly

13:03all of you.

13:04Yes indeed.

13:05You will discover

13:05in this video

13:06that we are in the process

13:07to have a conversation,

13:09we think we're having a conversation

13:10with a machine

13:11that sometimes people

13:12think human,

13:13it's so well done.

13:14Yes,

13:15she even cheated

13:16the Turing test.

13:1798% of people today

13:19think they have business

13:19to a person,

13:21but it is above all

13:21a machine

13:22which defines words

13:23depending on the context.

13:24And this system

13:25will be able to define

13:26that in 885 messages

13:29that she analyzed

13:30of the last 81 weeks,

13:32they follow.

13:33What do I want?

13:34What do I like?

13:35What is the ideal woman?

13:36What do I want to do?

13:37with my money?

13:38All this,

13:38the model is capable

13:39to do it.

13:40What is my political vision?

13:41What do I like?

13:42What don't I like?

13:43Brief,

13:43I'm being watched

13:44by my own bot.

13:46Yes indeed.

13:46I'll tell you something.

13:47If it's free

13:48or it costs 17 euros,

13:50it is that the product,

13:51It's you.

13:52That's kind of the message

13:52that must be kept in mind.

13:54If you don't pay

13:55$250 a month,

13:57you are in the process

13:58to train the model.

13:59And it goes much further

14:00than that.

14:01Every time you sign up

14:03data,

14:03your health data,

14:04fitness,

14:05of finance,

14:06rental,

14:07uses,

14:08historical,

14:08where you are from,

14:09IP address,

14:10everything is collected

14:11to make your chatbot

14:13your best Pukav.

14:15You are in the presence

14:16of the largest Pukav.

14:17Yes indeed.

14:18This model,

14:19including certain models,

14:20are the worst

14:21and will bring up

14:22all data.

14:22SO,

14:23forget that France

14:24and Europe said

14:25we don't want it.

14:26They don't care.

14:27They do it anyway.

14:28This means two things.

14:29If you are a business

14:30and you have data,

14:32customers,

14:32chatbots,

14:33or you want to have it installed,

14:34you are able

14:35to do something super powerful.

14:37I am addressing people

14:38who make chatbots

14:38of the company.

14:39Today we can

14:40to classify

14:41of trend

14:42of your customers,

14:43of segmentation,

14:44to understand it.

14:45What is his psychological type?

14:46What is your customer avatar?

14:48Don't ask AI

14:49to do it by magic.

14:51Use the data

14:52that you have

14:53of your customers.

14:54Your emails,

14:55your exchanges,

14:56put a chatbot

14:57on the interface.

14:58You don't know how to do it?

14:59Contact me.

15:00You have everything in description.

15:02Afterwards,

15:03what does that mean

15:04for an average user?

15:06You know,

15:06you think you know it,

15:08but it goes much further

15:09than you think.

15:11Are you

15:11a risk to society?

15:13The model is capable

15:14to predict it.

15:15More precisely,

15:16it's not magic,

15:17how is the model,

15:18using words

15:19that you use,

15:21will define statistically

15:22in what type of profile

15:24you belong.

15:25And that's how it is

15:26that the model will proceed.

15:27But we don't give it to him

15:29the magic prompts.

15:31The magic prompts,

15:32It is

15:32“Hi, ChatGPT!”

15:34Put your magic word,

15:35put your magic word,

15:36and put your magic word,

15:37and you have the whole magic program

15:39which will appear.

15:40That,

15:40you must understand

15:42that this is the reason

15:42for which

15:43you were disappointed

15:43from TchatGPT.

15:45You simply discovered

15:46than by putting words

15:46in an interface,

15:48even three words,

15:49the model is capable

15:50to make a sentence

15:51and even a text.

15:52This is the very property of LLMs.

15:54They are word compilers.

15:56The question,

15:57What do you want to do?

15:58Do you want to work?

16:00Or,

16:01are you fed up

16:02to waste time

16:02with this type of unnecessary interaction?

16:05Unfortunately,

16:06the majority of entrepreneurs

16:06Today,

16:07apart from writing text

16:08for social networks,

16:09are not capable

16:10to use it.

16:11You are in the process

16:11to do Google BIS.

16:13Yet,

16:14you understood

16:14that you are sitting

16:15on a very powerful tool.

16:18Read what I say,

16:19it earns me the wrath

16:20of these influencers

16:21to 300,000 subscribers

16:22because they make money

16:25telling you

16:25roughly

16:26that with

16:27some 5-6 hours

16:29of generic, banal videos

16:31and products

16:33discounted

16:34at 80%,

16:35you have the skills

16:37of prompt engineering.

16:38Well no.

16:39Well no,

16:39you're not going to do much

16:40apart from texts

16:41copy-pasted

16:41as they do

16:42to say

16:43if the blog

16:44running

16:45it's cool or not cool.

16:46Today,

16:48you realize

16:48that you have in your hands

16:49a tool

16:49which is above all

16:50of data classification.

16:52So if you have

16:53millions of texts

16:54to classify,

16:54you need a data analyst.

16:56If you have

16:56a few hundred

16:57or thousands,

16:58with a little patience

16:59and the right strategy,

17:00we can arrive

17:01to classify your customers

17:03and what you understand

17:04is that you are

17:05the product

17:05of a chatbot

17:06which has in appearance

17:07summer free.

17:08And yes,

17:09how many of you

17:09would be ready

17:10to pay 250 euros per year?

17:12Do you pay them?

17:13Put it in the comments for me.

17:14Me neither.

17:15But there are strategies

17:16to protect your data

17:17and I'm talking about it of course

17:18in training.

17:19In training,

17:20I'm teaching you several things.

17:21This is not hype

17:22what must be done,

17:23it is knowing how to use

17:24the tools

17:25with the right strategies

17:26for the professional field.

17:28And I'll tell you why.

17:29I had someone

17:29who came to me

17:30that I know

17:30who told me

17:31I am in the middle of a depression,

17:32I have a super high score,

17:33close to 90 out of 100

17:35at the clinic

17:36and I tell him

17:37ah that's good,

17:38I have a prompt

17:39which makes that

17:39when you send it

17:40in memory

17:41from ChatGPT,

17:42you have a set

17:43of your conversations

17:44who can give you

17:45maybe an opinion

17:45roughly

17:47objective

17:48of your interactions.

17:50Are you

17:51the person also black,

17:53also depressed

17:53that you think you are?

17:54I tell you this

17:54because people

17:55today are in the process

17:56to confuse AI

17:57with a person.

17:58The border,

17:58I told you,

17:59she is thin.

18:00Do you,

18:01Does this happen to you, by the way?

18:02Do you think

18:02to be present

18:03of a person,

18:03of a friend?

18:04Finally I saw it

18:04on lots of forums,

18:05I am really disturbed

18:06by this syndrome

18:07people.

18:09Tell me in the comments.

18:10Are you in the presence

18:11in your opinion

18:12of a person or not?

18:13With ChatGPT.

18:13Is it

18:14your confidant?

18:15SO,

18:16the person helped himself

18:16so prompt

18:18that I gave you

18:19and realized

18:19of a thing,

18:20she was not

18:20the person also black

18:22that she claimed

18:22and that she thought she was

18:23when she was

18:24in consultation

18:25with the doctors.

18:27Because,

18:28of a probable reason,

18:29is that we put ourselves

18:30in character

18:31in which we are.

18:32And when you are

18:32with 800,

18:33900,

18:341000 messages

18:34with an AI,

18:35you realize

18:36of a thing,

18:37it is that she has

18:38an overview

18:38much larger.

18:39And the advantage,

18:41is that you have

18:42a look that is different.

18:43And the person came out

18:44in fact of the exchange

18:45with AI

18:45saying

18:46but in fact,

18:46I am full of other things

18:47that I no longer thought.

18:49I no longer had

18:49the ability to see it.

18:51And AI,

18:52thanks to this ability

18:53amazing,

18:55fix blocks

18:55of attention

18:56on the elements

18:56important

18:57has arrived

18:58to give an image

19:00different and positive

19:01to the person.

19:02And that's when I tell you,

19:03you are a therapist,

19:04you are a coach,

19:05you are a doctor,

19:06you pass by

19:06of an incredible tool.

19:09No, not incredible.

19:09because he will do

19:10your job for you,

19:11but because using

19:12of the right method,

19:13of the right strategy,

19:14This tool will allow you

19:16to accompany you

19:17in decision making

19:18because it will allow you

19:20to pay attention

19:21over millions of hours

19:22of exchanges

19:23and conversations

19:23audio

19:24or meetings

19:25and make appear

19:26points

19:27that your brain,

19:28we as humans,

19:29we can't keep

19:30all in memory

19:31while the model

19:32is capable,

19:33him,

19:33to do it.

19:34And that's when I say to myself,

19:35these attention mechanisms,

19:36you have to control them,

19:38not just anyone.

19:39I tell you,

19:39Today,

19:40It's sad to say.

19:41It bothers me.

19:42But the best way

19:43that you become aware of it

19:44is that you share

19:45this video

19:46to raise awareness

19:48to people

19:48that they were made to do

19:49Wrong Way

19:50on the use of models

19:51by making them believe

19:52that it was

19:53so simple

19:54what to ask for

19:55to Aladdin's genie

19:56all solutions

19:58to their problems

19:59and to have

20:00actually words.

20:01But these words

20:02are not the solution.

20:03These words,

20:04It's statistical prediction.

20:06And that,

20:07that's the reason

20:07for which

20:07you are disappointed

20:08and that we must change

20:09the strategy.

20:10So I will give you

20:11the prompt I used

20:12which actually allows

20:13to go down

20:14in the interior

20:15of memory

20:16of ChatGPT.

20:17It allows like this,

20:18if you are a business

20:19or a person

20:20you realize

20:20of all finesse

20:22what the model can have

20:23to classify

20:24your discussions,

20:25your psychological type,

20:27the type of decision

20:28that you take.

20:29Are you guided?

20:30by emotions?

20:31Do you have

20:31beliefs?

20:33How do you see yourself?

20:34yourself ?

20:34Do you have

20:35psychological motivations?

20:37You have phobias,

20:38fears,

20:39products

20:39that you would like to buy?

20:41Brief,

20:41you are in the presence

20:42of the best sales system

20:43who knows you best

20:44than yourself

20:45and which can be used

20:46for a company

20:47to extract data.

20:48SO,

20:49you understood it,

20:50entrepreneur,

20:51therapist

20:52or customer,

20:53everyone

20:54won

20:55to understand

20:55the tool

20:56which is neither all black

20:57nor all white,

20:58but you have to understand

20:59the consequences,

21:01how to use it

21:01and how maybe

21:03not to take oneself

21:04the feet

21:05with the tool

21:06who can tomorrow

21:08become your worst snitch,

21:10your poucave,

21:11the one who will say

21:11to your employer

21:12he didn't pass

21:13so many hours

21:14that he claims

21:15on the file,

21:17he went on the internet

21:18look at the results

21:20of the football match.

21:21Yes,

21:21the model,

21:21it is capable

21:22to be able to do

21:23this kind of thing

21:23and even worse

21:24because as he has access

21:25now to your mailbox,

21:26he can absolutely say

21:27he writes to Justine,

21:30it may be

21:30discussions

21:32who are outside the professional framework

21:33and if your employer

21:35or your wife

21:36receives the email

21:37because yes,

21:37he has access to your email

21:38NOW,

21:39you did

21:40the blunder

21:41all alone

21:42simply by knowing

21:43that this model

21:44is capable

21:44to make decisions.

21:45So no,

21:46It's not to sell you

21:47of the hype,

21:49it's to make you

21:49become aware

21:50the way to interact.

21:52You are not going to

21:52get there

21:54by telling the taxman

21:55it's me

21:56who tried

21:56to optimize my company.

21:58Don't be

21:59the product

22:00of your chatbot.

22:02If you want to know more,

22:03you have the training

22:04in description,

22:05subscribe

22:06on the chain,

22:06it's not already done,

22:07Above all, share.

22:08For what ?

22:09Because it's time

22:10to stop

22:11to lose

22:12of time,

22:13I tell you again,

22:14three years of life,

22:15no chatbot

22:16will not give them back to you.

22:17For information

22:18without filter

22:19on AI,

22:20This is where it happens

22:21and if you want to do

22:22a video

22:23with me,

22:24a subject,

22:25you want to propose

22:25your skills,

22:27send me a DM,

22:28we talk about it,

22:29we'll organize it,

22:29It will be with pleasure

22:30to debate it.

22:31You don't agree?

22:32Why not ?

22:33You tell me.

22:34SO,

22:34I am open

22:35to the discussion

22:36but there is one thing,

22:38we can't say

22:39to people one thing,

22:40in my opinion,

22:41It's deceiving them.

22:42Make them believe

22:43that everything is simple

22:44and easy

22:44and that upon simple request,

22:47AI will do it

22:48in your place,

22:49by making believe

22:49that it is

22:50of skills,

22:51that's an argument

22:52that I will refute

22:54because that's how it is

22:55that we deceive

22:56of millions of people

22:57who do not have the skills

22:58and who did not have

22:59the opportunity

23:00to have something else

23:02that this type of argument

23:03and that unfortunately,

23:04I find that it is

23:05deceptive marketing.

23:06Tell me what you think

23:07in comment,

23:09I would be delighted

23:10to know your point of view,

23:11aren't we fed up?

23:12Today

23:12to introduce ourselves

23:13miracle solutions

23:14which are not miracles at all?

23:16I tell you,

23:17it's now

23:18that you will have the prompt

23:19to access the information.

23:21I'll put you in touch

23:21in the description

23:22a small button to click

23:23which will allow you

23:24to access this information.

23:26Comment me

23:27the word prompt

23:29and I will give you the link

23:30to be able to access

23:31to the memory hack.

23:33Disclaimer,

23:34don't forget

23:35than what NIA tells you

23:36These are statistical probabilities.

23:38You are not

23:38presence of a specialist

23:41and if you need help,

23:42call a professional.

GPT‑5 se souvient de tout (et ça fait peur…)  #chatgpt #opeani #gpt5

Catégorie

Transcription

Écris le tout premier commentaire

Recommandations

GPT‑5 se souvient de tout (et ça fait peur…) #chatgpt #opeani #gpt5

Catégorie

Transcription

Écris le tout premier commentaire

Recommandations

GPT‑5 se souvient de tout (et ça fait peur…)  #chatgpt #opeani #gpt5