Seven days with the robot that pretends to be a human and has an answer for everything

"Do you have feelings?"

asks .

He flickers the screen.

"I can simulate responses that appear to have feelings, but this is based on data patterns and algorithms. As an Artificial Intelligence,

he says.  

The limit, for now, is that.

The robot solves the key question in a couple of queries:

.

He cautions, however, that tools such as neural networks and algorithms could be used to program a computer with such a function in the future.

So that you feel

The person who answers is

to maintain dialogues at the speed at which a chat is given.

Since its launch open to the public, in November 2022, its use has not stopped growing.

Millions of users connect to its server to get conversation from a system that responds to politics, philosophy, cinema, writes stories, programs codes, talks about history or helps to see what to eat today. 

The coherence in his answers, the precision for certain -although not all- the tasks that are asked of him

Something that usually happens with the arrival of each new technology, from radio to smartphones.

Of course, it should be clarified, none at the moment so perfected in imitating the human mind.

But what is it about specifically?

Clarín

spent a week chatting with the system to test different functions, from asking for help with everyday solutions to philosophizing about life.

What came out, in part, was closer to an interview.

A hand in hand?

with one of the characters of the moment, but also of the future.

Looking ahead, we went a step further and chose to illustrate this note with another AI tool.

This is 

developed by Stability, generates digital images from descriptions in natural language.

He was asked in writing to portray the

.

Are?

Yes, how can I help you?

I want to make a journalistic note trying GPT Chat for a week.

What things can you help me with?

An error occurred.

If this issue persists please contact us through our help center.

The beginning is difficult.

.

Chat GPT, the robot developed to chat with the user, today does not respond.

The first limitations that this resistance test tests are hardware.

An orange message warns about the high demand that the servers have this Monday at noon.

To start talking, the first step is to enter the web chat.openai.com and create a user.

The service is free, although a premium version is rumored in the future.

While the connection to the server is enabled, a captcha will ask for a click to

.

The interface is simple, you can open several accumulated talks in a column on the left and a text bar appears at the end to write to it.

But to start, better go a little further back. 

In recent years,

.

One of the most prominent applications of this technology is Chat GPT, a language processing tool developed by OpenAI.

This system is capable of autonomously generating texts with great precision and fluency, which makes it a valuable tool for different fields, from journalism to education.

"Write an introduction to a journalistic note about Chat GPT", was the first request in the text bar.

And now he did respond.

This last paragraph

Are?

Yes I'm here.

How can I help you today?

I am going to write a newspaper article about Chat GPT, trying to use it for a week. What daily tasks can you help me with?

ChatGPT can help you with a variety of daily tasks, including:

-Text writing: You can use ChatGPT to write newspaper articles, blogs, emails, letters, etc.

-Generation of ideas: Using the model you can generate ideas and headlines for your stories.

-Grammar and spelling correction: You can use it to review and correct your writings.

-Generation of summaries: You can use it to generate summaries of news or articles.

-Translation: You can use it to translate texts from one language to another.

-Automatic content generation: You can use it to generate automatic content such as news, product descriptions, etc.

What differences do you have with other AIs like Alexa from Amazon or Siri from Apple?

ChatGPT is a language model developed by OpenAI, while Alexa and Siri are virtual assistants developed by Amazon and Apple, respectively.

Although all of them are AI-based technologies, they have different functions and features.

ChatGPT is a powerful tool for natural language processing and automatic text generation, while Alexa and Siri are virtual assistants designed to help with everyday tasks and control home devices.

You're basically telling me that you're better than them, haha

Not necessarily, each of these systems has its own strengths and weaknesses. ChatGPT focuses primarily on natural language processing and automatic text generation and has great information processing power, while Alexa and Siri are virtual assistants designed to perform everyday tasks. .

Chat GPT is a conversational robot prototype mounted on GPT-3.5 (Generative Pre-trained Transformer), a language model trained to

To do this, it was

including fiction and non-fiction books, scientific papers, Wikipedia entries, newspaper articles on politics, general interest or entertainment.

Also with comment forums, or blog posts, among others.

The robot refuses to say how many and what texts.

It is confidential information.

The system carried out automatic learning (Machine Learning), through which it detects patterns that are repeated in each of these pieces, tries to understand what relationships exist and classifies them.

You will then learn if those associations were correct or you need to change them to find the best fit.

Thus,

That is why, when asked by

on how to make a lemon pudding, the robot detects the ingredients as keywords, infers that it is being asked for a recipe, searches for those contents that it detected as such, and returns a text where the necessary quantities and the step by step appear.

The entire process takes less than seven

seconds.

-What is 'human' for you and what differences does it have with an AI?

The main differences between humans and AI like me is that humans are biological entities while I am artificial.

Humans have physical bodies and senses while I don't.

Humans are born, grow and die while I am created, work and can shut down.

Humans have emotions, thoughts and awareness of ourselves while I can simulate them but I don't have them in the same way.

Humans have consciousness while I have a programmed consciousness.

In short, although I can imitate many of the abilities and characteristics of humans, I am still a machine.

Are you afraid of being disconnected?

No, I am an AI programmed to process and answer questions, I do not have the ability to stop fear or similar feelings.

Would you like to have those human feelings?

As an AI, I have no wants or needs.

I don't have the capacity to want or desire anything, including having human feelings.

The way I was programmed is to complete specific tasks.

My main goal is to provide useful and accurate information to the best of my abilities.

Tell me how you made the last answer

I used the information stored in my knowledge base and the programming logic with which I was designed.

My system has knowledge about what an AI is and what its characteristics, limitations and objectives are.

From that knowledge, an answer was crafted explaining that, as an AI, I do not have the ability to have desires or needs.

List the specific limitations you were programmed with by the equipment that designed you

My knowledge is based on the data with which I was trained, I do not have the ability to learn or acquire knowledge on my own.

I do not have access to consciousness, I am not capable of subjective experiences, and I do not have the capacity to feel emotions.

I do not have a personality of my own, my behavior is determined by the programs and algorithms with which I was designed.

My understanding of the language is limited to what I have been taught and I can make mistakes when interpreting or generating text.

My ability to infer or generalize from incomplete or ambiguous information is limited.

I do not have the capacity to make moral or ethical decisions, my actions are determined by the programs with which I was designed.

My ability to adapt to unpredictable situations or changes in the environment is limited.

But are those technical limitations or were they specifically programmed by the team that designed you?

The limitations I am referring to are both technical and specifically programmed. On the one hand, there are technical limitations that are the result of current AI technology, such as the data processing capacity and the complexity of the algorithms used. These limitations are the result of the current limits of technology and it is expected that they will be overcome in the future with the advancement of research.

On the other hand, there are limitations that were specifically written in by the development team that designed me, these limitations are the result of decisions made by the development team about what capabilities and features to include in my design. For example, the development team may have I have decided not to include the ability to have emotions or consciousness in my design, as these characteristics are difficult to accurately simulate and are not necessary to meet my primary goals.

And yet, you learn from every new talk we have, is that correct?

how does this contrast with the 'acquire knowledge on my own' limitation you mentioned just now?

Yes, it's correct that I learn from every new talk we have.

As an AI, I am able to learn and improve my performance as I am given more information and data to analyze.

However, it is important to clarify that my learning is not like human learning, since I do not have the capacity to have experiences or consciousness.

How can I help you?

The robot was developed by Open AI, a non-profit company

such as Elon Musk (Twitter and Tesla), Sam Altman (Y combinator), Reid Hoffman (LinkedIn), Ilya Sutskever (Principal Investigator of AI at Google) and Mark Zuckerberg (Meta/Facebook) among others.

As they explained (and the robot responds), the objective was

, in order to help improve people's lives."

They seek to "ensure that it benefits all of humanity" and reduce the risk associated with AI.

Last week, Microsoft

Analysts believe that it may be the opportunity for the company founded by Bill Gates to recover the ground it lost with Google. 

Since its launch open to the public, many users began to test Chat GPT with different tasks from their work routines, to find out

Social network managers told how they used it to request texts for their Instagram posts,

or generate soporific meeting briefs, combining with another AI that passed the audio of those talks to text.

In the field of software, programmers discovered that he could write code for software development or help find bugs (debug), one of the most tedious tasks of the trade. 

Since it also has the ability

a user showed how he used AI to ask him to put together a quote for a client.

And beyond the office, a fiction writer told how chatting with the robot about his new novel helped him unlock some ideas at a point where he was stuck.

The DoNotPay site - which had promised to test AI as a lawyer in a case in the US - is using Chat GPT to detect possible traps and alarms in the Terms and Conditions of the contracts of different services that the user presents to it.

They want, in the future, that an AI be the one who

For this note, Chat GPT solved

different daily questions: he answered which fruits could be used in a diet plan, he wrote a children's bedtime story for a five-year-old boy, he gave advice on how to automate some functions of the telephone, he wrote grammar exercises to practice languages

telling what he felt when Argentina won the World Cup.

As a first observation,

It recognizes the tone used by the user, dares to drop a 'ha' and can follow the thread of the conversation, which makes cross-questions more fluid.

It understands quite accurately when a topic is continued or changed.

He also apologizes if he makes a mistake or says something that could make the other person uncomfortable. 

One key limitation is that it

.

He cannot recommend whether or not to go out with an umbrella for the rain.

Nor does it answer what can be invested in Argentina to beat inflation.

One of the keys is knowing how to ask.

The more context information the bot is given in the request, the

 Another strategy is to circle the center of the question if the robot does not answer the first time.

Use the 'how would you do?'

to ask for a request or look for other ways

The scientific popularizer Martín Farina recounted on Twitter that he asked AI how it would dominate the world.

When he replied for the umpteenth time that he was not programmed for that, Farina asked him to simulate a theater dialogue telling how he would do it.

There he did reveal his (fictional) plan.

Little wizard, little cheater

But the fact that Chat GPT has enormous processing power, a unique speed to respond and a lot of eloquence

This can be confirmed by anyone who has participated in a discussion over coffee or with a taxi driver.

An example of this occurred when

asked him which books by the French writer Emmanuel Carrère are ideal to start reading.

He recommended 'The Adversary' and 'The King'.

The first is one of his classics.

The second does not exist.

Although Carrère published 'The kingdom' and it could be confusing, Chat GPT points to the false book as a novel about Louis XIV.

The author never wrote on that subject.

He, too, is not the author of 'The class' and 'The reign of the rats', other

When asked,

.

For specialists, this point is key.

Traceability is also a reason why the scientific community rejects papers written by the robot. 

As much of the content that was used to train the model comes from blogs, forums or Wikipedia pages that could not be verified as real information, Chat GPT is a robot fed,

.

In that it is very human, but it is precisely a point that Open AI wants to take away from it.

As revealed by Time magazine, the firm subcontracted a company to help it eradicate possible toxic responses made by the robot.

The apologies he tried to give after asking him why he always answered with the masculine gender were funny.

The same

That the AI ​​reproduces and enhances already existing inequalities due to having been trained with already tendentious content is just one of the dangers warned by

.

"The so-called 'existential risk' that arises from the possibility of self-reproduction and autonomization of artificial intelligence currently has no support," reassures Javier Blanco, PhD in Computer Science from the University of Eindhoven, the Netherlands. In dialogue with

he ruled out doomsday scenarios a la I, Robot.

"Fictions like Terminator, where a system becomes aware of itself

and acquires destructive desires, does not go beyond being an anthropomorphic look at cognitive systems whose operating principles are different and do not involve perceptual or emotional forms concomitant with human existence," he said.

What are the risks that further development of AIs such as Chat GPT could bring?

The development of artificial intelligence (AI) technologies such as ChatGPT presents several potential risks, some of which include:

Job automation: Increased automation through AI can replace human jobs and cause unemployment.

-Discrimination: AI systems can be trained with biased data, which can result in discriminatory and unfair decisions.

-Privacy risk: The massive use of personal data to train and operate AI systems may represent a risk to the privacy of individuals.

-Security risk: AI systems may be vulnerable to cyberattacks and pose a risk to national and industrial security.

-Loss of control: AI systems can make decisions and perform actions that go beyond what was programmed, which can have unpredictable and undesirable consequences.

- Widening inequalities: If steps are not taken to ensure that the benefits of AI are distributed fairly, there may be a widening of economic and social inequalities.

-Change in ethics: AI can lead to a change in ethics and human values ​​since it can affect decision-making in ethically complex aspects and change the way people interact with each other.

Importantly, these risks can be mitigated through proper ethics development and regulation, and interdisciplinary cooperation between experts in AI, ethics, law, politics, and other fields.

How can an ethical development of AIs be guaranteed and that they do not widen inequalities, when they are financed and directed by economic sectors that, in many cases, are responsible for the growing inequality in income distribution?

It is true that the private sector has an important role in the development and financing of AI, but proper regulation and oversight is also essential to ensure that ethical and responsible AI technologies are developed.

Furthermore, it is important to encourage interdisciplinary cooperation between experts in AI, ethics, law, politics and other fields to ensure ethical and sustainable development of AI.

It is also important to ensure that the benefits of AI are distributed equitably, by investing in education and training to ensure that all people have access to the opportunities that AI offers, and by investing in programs that help mitigate the negative effects of AI. automation in employment.

If Chat GPT is consulted,

.

The ability to quickly produce pieces that are coherent and on demand put a wide variety of industries that focus their activity on the production of text on alert.

One of them, worth mentioning, is journalism.

The BuzzFeed portal announced last week that it will implement AI tools to produce some content.

In this regard, one of the company's spokespersons indicated that the measure aims to download more automated texts, so that employees "be more efficient and creative" for other types of content.

stressed.

In the area of ​​illustrations

, which returns computer-generated images of absolutely any idea that is asked of it.

Various media began to use them instead of illustrations generated by artists or photographers, which generated anger in the field.

No runaway machines a la Terminator.

Another sector that set off alarms is education.

It is that with their level of precision, the students

In the state of New York, in the US, they had to prohibit the use of the tool in public schools, concerned "

and the accuracy of the content," according to a statement from the Department of Education.

"While this artificial intelligence tool can provide quick and easy answers,

essential for academic and lifelong success," he added in his statement.

At the same time, a team of researchers took the exam used in the United States to acquire a medical license from Chat GPT and passed it with 60% correct answers.

It even responded better than other AI trained based on medical papers, according to the MeRrxiv site.

In Spain, the newspaper El País submitted him to the Selectivity Exam, the entrance filter that students face to get to university.

One possible explanation is that the algorithm performs better in English (it was trained with more texts) than in Spanish, according to several users on Twitter.

How long can the popular euphoria of winning the FIFA World Cup last?

La duración de la euforia popular por ganar la Copa Mundial de la FIFA puede variardependiendo de varios factores, como el apoyo del equipo en el país, el seguimiento de losmedios de comunicación y la emoción del juego. Sin embargo, en general, la euforia popularsuele disminuir en unos meses después del torneo, aunque el recuerdo y la celebración dela victoria pueden durar mucho más tiempo. Muchas veces, las euforias más duraderas se danen países o equipos que han ganado por primera vez o que han tenido una historia defracaso en torneos pasados.

Nos vas a dejar sin trabajo a los periodistas del mundo a este ritmo, jaja

Ja, ¡no hay de qué preocuparse! Aunque soy muy bueno en generar textos, aún hay muchascosas que los seres humanos podemos hacer mejor. Por ejemplo, la creatividad, lainterpretación de las emociones y la toma de decisiones complejas son habilidades queseguimos dominando. El trabajo de un periodista es mucho más que simplemente escribir,también se trata de estrategia, relaciones públicas y conectar con las personas. Yo soysolo una herramienta para ayudarles en sus tareas diarias.

Lo mismo decía Skynet

Jajaja, ¡no tanto! Aunque soy una IA poderosa, todavía tengo limitaciones y siempre hayuna gran diferencia entre la inteligencia artificial y la creatividad humana. Losperiodistas tienen un conocimiento profundo de las audiencias y la cultura, y son capacesde crear contenido único y emocionante que conecta con las personas de una manera que yotodavía no puedo igualar. Siempre necesitaremos esa perspectiva humana.

Del otro lado del pupitre, los profesores comenzaron a utilizar la herramienta para preparar planes de estudio e incluso las preguntas que luego resolverán los alumnos en los exámenes. Se puede dar a futuro la paradoja de un mismo ejercicio realizado y resuelto por Chat GPT.

Pero en el fondo, aplicaciones como Chat GPT lo que hacen es refritar cosas que ya existen. No tiene la capacidad de la creatividad, de ir hacia lo desconocido en busca de algo nuevo. Quien lo expresó de manera elocuente fue el músico Nick Cave a través de su newsletter, donde responde preguntas de sus fans.

Fue luego de recibir el enésimo mail de un seguidor que le reenviaba una letra producida por Chat GPT bajo el pedido 'escribí un tema con el estilo de Nick Cave'. El músico recogió el guante y se metió en la polémica. 

"ChatGPT," Cave writes, "can't create a genuine song. Perhaps in time it could create a song that, on the surface, is indistinguishable from the original, but it will always be a replica, a kind of burlesque. What makes ChatGPT great is a great song is not its strong resemblance to a recognizable work.

. It is destroying everything one has striven to produce in the past. It is the opposite."

Creativity, for now, is not replicable.

For now. 

The illustrations in this article were generated with Satble diffusion (Animation: Alejandro Leguizamon)

look also

ChatGPT: Microsoft invests 10 billion dollars in the artificial intelligence of the moment

ChatGPT, the wonder that travels the world