Inteligência Artificial

Botando uns pingos nuns is…
(English version below)

O que acham que é?

ChatGPT e afins.

O que é?

Um campo de estudo sobre métodos para fazer com que o computador tome decisões “espertas” com mínima ou nenhuma intervenção humana.

O chatGPT é uma ferramenta que foi construída com um desses métodos. Por ela ser fácil de acessar e usar, e pela qualidade das respostas, é uma ferramenta que se popularizou muito. Mas ela é uma de muitas outras ferramentas usando um de muitos outros métodos da área de inteligência artificial.

O chatGPT é inteligente mesmo?

Não. O chatGPT é um jogo de probabilidades. Vou tentar explicar com uma analogia.

Imagina que você memorizou todas as obras da grécia antiga, em grego, sem saber falar grego. Ou seja, você sabe todos os símbolos, em todos os documentos, e a ordem que eles aparecem, mas não faz a menor idéia do que eles significam.

Aí alguém escreve pra você uma pergunta em grego antigo. Você vê aqueles símbolos e, sem saber o que significa, vai procurar em todos os outros símbolos que você memorizou se tem alguma coisa parecida. Aí você acha umas coisas similares e começa a construir uma reposta. Primeiro escolhe uma palavra que geralmente segue aquela pergunta, ou perguntas muito parecidas (porque têm muitos símbolos em comum). Daí você dá uma pesquisada no que geralmente segue aquela palavra, e escolhe a segunda aleatoriamente dentre os candidatos mais prováveis, e assim por diante.

É uma idéia simples, implementada muito complexamente com várias transformações de símbolos pra números (porque é mais fácil pro computador trabalhar com números), e que funciona incrivelmente bem. Mas que, na minha opinião, diz mais sobre a nossa regularidade e padrões existentes em tudo que a gente produz do que sobre um computador sendo mesmo inteligente.

É perigoso?

A tecnologia por ela mesma, não. A tecnologia na nossa mão descuidada, provavelmente sim. Mas essa é a história da humanidade. As tecnologias surgem, todo mundo fica empolgado e espreme tudo que pode delas até as coisas começarem a dar errado, e aí a gente pensa como consertar. Foi assim com a agricultura, com os carros, com a internet (que ainda não foi consertada, mas já deu ruim), e vai ser assim com a inteligência artificial. Tem potencial pra dar certo, e potencial pra dar errado, mas não temos o tempo (ou talvez a calma) pra refletir antes da coisa se popularizar e ser usada pra diversas tarefas, boas e más.


Artificial Intelligence

Crossing some t’s and dotting some i’s…
(Versão em português acima)

What do people think it is?

ChatGPT and family.

What is it?

It is a field of study about methods for computers to make “smart” decisions more or less autonomously.

ChatGPT is a tool built using one of those methods. As it is easy to access and use, and for the quality of answers, it has become widely popular. But it is one of many tools based on one of many methods from artificial intelligence.

Is chatGPT really intelligent?

No. ChatGPT is a probability game. I will try to explain using an analogy.

Imagine that you have memorized all the works from ancient Greece, in Greek, without knowing Greek. That is, you know all the symbols, in all documents, and the order in which they appear, but you have absolutely no idea what they mean.

Then someone sends you a question written in ancient Greek. You see those symbols and, without knowing their meaning, look for similar sequences of symbols among the ones that you have memorized. You find some similar stuff and start building an answer. First you choose a word that typically follows the question symbols or sequences like it. Then you search what usually follows that word, and choose a second one randomly among the more likely candidates, and so on and so forth.

It is a simple idea, implemented in a very obscure way with many transformations from symbols to numbers (because it is easier for computers to work with numbers), and that works amazingly well. However, in my opinion, this says more about our regularity and patterns existing in everything we have produced, than about a computer being actually intelligent.

Is it dangerous?

The technology in itself, no. The technology in our careless hands, probably yes. But this is the history of mankind. Technologies arise, everyone gets excited about it and use it to its limit until things start to go wrong, and then we think how to fix it. It was like this with agriculture, cars, internet (which we did not fix yet, but it already went wrong), and it will be like this with artificial intelligence. It has the potential of being a good or a bad thing, but we do not have the time (or perhaps the patience) to reflect before it becomes too popular and used for all sorts of things, good and bad.


P.S.: AI likes my post about AI 😂

My experience in a male dominated discipline

I was asked to participate on a panel at the university about diversity, and my task was to talk about “my experience in a male dominated discipline”. I kept thinking about this for the last couple of days, and I am always unsure on how to approach this topic. On the one hand, I have never felt I was at a disadvantage in my career due to my gender. On the other hand, I know of many women who do struggle in their workplace because of their gender. But I was asked to talk about *my* experience, so here we go.

Throughout my career I have met wonderful people (men and women) whom I get along with and who never made me feel for a second that I was “different”. So many names just pop up in my head and I feel incredibly fortunate for that. Most of the times our commonalities are much greater than our differences, and I like that none of these people seem to care that I am of a different gender, or that I look differently. I was very lucky to have been able to choose who I work with most of the time (academia perk?), and looking back I have surrounded myself with colleagues and friends that made working a pleasant experience. I do not remember many instances where I needed to “put up with someone” for too long. Perhaps a consequence of that is that I hadn’t really paid so much attention to the whole gender biased business until much later in life.

As I became more aware of gender differences, I started noticing certain things here and there, and listening to other people’s experiences. Today, my opinion is that the core of the problem is beyond gender, and probably much more related to personality. We live in a society that values the outspoken, the ambitious, the confident one. We are easily impressed and convinced by those that speak their messages loud and clear, even if they are not sure of what they are talking about. And we feel that cautious and nuanced messages are from insecure and unsure people. We are drawn to certainty, even when it is not certain at all. That leaves the introspective and careful people at a disadvantage. (Ironically, the more nuanced messages are usually the more well-thought ones). The fact that women are generally raised to be more agreeable, and less combative, is incidental, and makes the problem worse for us, typically but not exclusively. However, being a cautious person in an overly confident world is tough for any gender.

So I feel like my struggles have been caused more because I am cautious and I don’t like to rush into things, than for being a woman. The way I found around this is to move closer to people that appreciate this trait, and move away from those that feel this is a weakness. I know I am lucky to be able to make this choice, and I know many that are not.

Adobe and Microsoft

Just got a message from the IT department at work:

Reports have emerged overnight that an Adobe Flash security flaw is being actively used across the Internet to attack and compromise computer systems. The Flash vulnerability affects Windows, Mac, and Linux systems but the exploit is targeting Windows at this time.

The attack comes in the form of malicious Flash content embedded inside an Office document (Word, Excel, PowerPoint) that triggers the Flash security flaw to compromise the computer.

I am laughing inside xD