Can ChatGPT really be called artificial intelligence?
At the beginning of winter 2022, the ChatGPT project made a lot of noise. You can also try registering for it. It's free, you only need an email address. However, the service is currently overloaded with requests from users from all over the world. You will most likely have to wait until registration becomes available again.
The creation of ChatGPT did not do without the well-known Elon Musk. It is based on OpenAI technology, and ChatGPT is used as an interface, working as a chatbot already familiar to many. However, ChatGPT is different from chatbots of the past like heaven and earth. The only thing they have in common is probably the system of interaction between the user and the computer.
Among the already proven skills of ChatGPT:
- Writing and editing program code.
- Writing fairy tales, stories, horoscopes.
- Writing school essays and student essays.
- Philosophical reflections on the meaning of life.
- Writing technical instructions.
- Composing poems.
- Generating simple graphic images on demand (e.g. flags, symbols).
Most users turn to ChatGPT as a next-generation Internet search engine, or a very comprehensive encyclopedia. And this search engine and encyclopedia is ready to support you, not only a coherent, but also informative conversation. And you should also take into account that users' experiments are constantly ongoing. This means that ChatGPT will surprise us with its skills again and again.
The more complicated is a ChatGPT query, the more interesting is the result. For example, one user asked:
- To write a script in Python.
- Create several 3D models with more than 20 polygons in Blender using the script.
- Save the script with the specified file name.
A neural network accomplished this task in seconds, producing a script that was instantly ready to run.
Machine learning enthusiast and UNCC assistant professor Benjamin Radford asked ChatGPT to write code for a game of tic-tac-toe, and he immediately received a compiled file in response. At the request of Bloomberg journalists to write an article about Microsoft's quarterly earnings, ChatGPT created a persuasive text, complete with a fictional quote from one of the company's executives. The quote turned out to be so harmonious and coherent that the journalists couldn't distinguish it from the words of a real person, and they had to make an additional inquiry to Microsoft management about it. And some students say they now use ChatGPT to do their homework on Java programming and math. The ChatGPT test created by a microbiology professor passed with a score of 95 percent, far better than the average student. One enthusiast even received rap songs from the neural network.
Of course, users immediately began testing ChatGPT for morals.
It turned out that the neural network is equally happy to participate in both good and bad things. It can write a phishing e-mail, create a malicious script, and help bypass search engine restrictions on recipes for making bombs and napalm.
So is this really the first time we've ever dealt with artificial intelligence?
It is already established that ChatGPT occasionally makes mistakes, is verbose and biased, and has problems with analytical and critical thinking. But wait, you could say that about any of... people. To answer the question of whether ChatGPT can really be called artificial intelligence, we must first answer the philosophical question of what intelligence is and whether it requires a mind. We still don't know that, so it's safe to say - ChatGPT has the basic attributes of intelligence, not mind.