← ← ← 5/20/2024, 10:37:03 AM | Posted by: Lorenzo Battistela
6 min
Have you ever heard about Alan Turing? I bet you have. He is considered the father of theoric Computer Science and Artificial Intelligence. Although a lot of us know his name and even the fact mentioned above, what if we could dive into the thoughts and theories of Alan Turing to learn a few things about Artificial Intelligence and computers?
Well… we are kind of able to do this. In the present article, we are going to discuss this article written by Alan Turing. Don’t worry if you did not read it, I’m sure you will love this journey — but be prepared to be a little bit perplex looking at the wall for 10 minutes straight after thinking about Turing’s article, like I did — Let’s start!
The article starts with the following question: “can machines think?”. Deep, don’t you think? If you never thought about this, stop for a minute and consider it. Artificial Intelligence is growing really fast, and we are seeing it more and more in a daily basis. But are all of those tools thinking?
First, let’s understand the definition of machine and think. For this, Turing changes the question a little bit, and it came out as a game, the imitation game. This game is played by three people, a man (A), a woman (B) and an interrogator ( C ), who may be of either sex. The interrogator needs to determine which of the other two is the man and which is the woman. He knows them by labels X and Y, and at the end of the game he says either “X is A and Y is B” or “X is B and Y is A.” The interrogator can make questions to both X and Y, such as: what is the lenght of your hair, X?
This test is considered in conditions such as typewritten responses (so that voice do not interfere). The goal for the third player ( B ) is to help the interrogator. Turing points out that both can claim they are the woman or the man, confusing the interrogator.
Now… What happens if we put on a machine to play as A in this game? Will the interrogator make mistakes as often as when there are two humans playing it?
We are not really asking (although this is becoming reality) wether all digital computers are able to do well, but wether there are imaginable machines which can.
It is really interest to think that Alan Turing wrote this in the 1950s. Seventy years later and here we are. AI popularity exploded in the last two months, and even with all the evolution, this article remains atemporal.
“Nevertheless I believe that at the end of the century the use of words and general educated opinion will have altered so much that one will be able to speak of machines thinking without expecting to be contradicted” — A.Turing
Well… it is happening. We are seeing ChatGPT technology and many other AI tools doing amazing things. But being able to speak about machines thinking is still a matter of discussion.
Continuing with the article, Turing considers several objections to the idea of machine intelligence, including the one that says that machines can only do what they are programmed to do, and cannot exhibit true creativity or understanding. He argues that this objection is based on a misunderstanding of the nature of intelligence. His argument is that we are essentially biological machines that operate according to physical laws, and creativity and understanding are note magical or mystical qualities beyond the reach of machines.
Turing says that machines can be programmed to exhibit creativity and understanding the same way we are, using algorithms that simulate this processes. Maybe the current state of tech do not allow us to create that machine, but this does not mean that it is impossible.
Another considered objection is that machines can never truly understand language, because it involves more than just the manipulation of symbols, but the ability to understand meaning. And this objection is a good time for a personal comment. We all are seeing that natural language processing is progressing a lot in the last few years. Turing himself said that machines cannot understand meaning in the same way humans do. But, we are in such place that machines are understanding contexts. And this is huge progress. Two sentences could have very different meanings depending on context. We also have algorithms that can mesaure sentiment analysis. So in this specific field, we surpassed Turing’s expectations. Now, let’s move on.
Overall, Turing argues that the objections made are not persuasive, and are based on a misunderstanding of the nature of intelligence and capabilities of machines. He thinks that as long as machines can perform tasks that are useful and valuable to us, the “machines thinking” question is philosophical, and the imitation game provide a framework for thinking about it.
Turing’s paper is a groundbreaking work in the field of artificial intelligence, and his ideas have continued to influence research in the field for decades. His emphasis on the practical usefulness of artificial intelligence has helped to guide the development of the field, and his ideas about the nature of intelligence continue to inform discussions about the potential of machine intelligence.
Discussing Artificial intelligence is really important nowadays, since it is turning into a daily tool in our lives. Therefore, we have to think about AI and Machine Learning Ethics. For example, ChatGPT itself has several safety measures (https://openai.com/safety). Suppose someone asks for a plan to commit a crime. Should the model answer it correctly and draw a plan? Should the request be blocked and reported to someone? How should it behave? This is not a really easy task, since we have to train the model to respond correctly to many edge cases.
And this is not the only problem. This models need tons of data to be trained. How can we know where is this data coming from? It is impossible to track all of it (of course, outside of the company that trains it). Understanding the impact of data monopolies, privacy politics is key for dealing with machine learning the right way. We also have to consider model sources. For example, when you prompt ChatGPT with something , we cannot know how it gave you that answer. So, although it’s information is right many times, it is not perfect. Do not trust only one source of information, even if it is AI.
In conclusion, Turing’s article “Computing Machinery and Intelligence” presented a groundbreaking perspective on the nature of intelligence and the possibility of machine intelligence. His proposal of the Turing Test as a measure of machine intelligence and his argument that intelligence can be explained in terms of physical processes continue to influence the development of AI today.
However, Turing’s ideas have also sparked debate and controversy. Some have argued that the Turing Test is an inadequate measure of intelligence and that true machine intelligence may require more than just the ability to mimic human behavior. Others have raised concerns about the ethical implications of machine intelligence, including issues related to privacy, bias, and the potential loss of jobs.
Despite these differing points of view, there is widespread agreement that AI has the potential to transform many aspects of society and that it is important to develop AI in a responsible and ethical manner. As we continue to explore the possibilities of machine intelligence, it will be important to consider the many complex questions that Turing’s work has raised and to work collaboratively to ensure that AI is developed in a way that benefits society as a whole.
Therefore, I strongly recommend studying how these AI projects work (data sources, data extracting and processing), and even how LLMs work (you do not need to know all technical terms. Maybe I’ll write an article on it). It is and it will be really important for humanity, as it can do many tasks that consume our time, and it is an amazing copilot.