There has recently been a huge technological revolution in artificial intelligence (AI) technology. The most important feature distinguishes this latest technology from its previous versions is its data processing capacity.
In fact, due to GPT-3’s high data processing capacity, it is not possible to understand whether the content is created by GPT-3 or a human. So, let’s look at what the GPT-3 is and its features.
Generative Pre-Trained Transformer 3 (shortly GPT-3) was developed by OpenAI, a San Francisco-based artificial intelligence research laboratory. It was introduced on May 14, 2020, it has been using as beta version since July, 2020 and it is still available for limited users. This technology, which is the 3rd generation of the GPT-n series, has 100 times more parameters than its previous model GPT-2. While the previous version has 1.542 billion parameters, GPT-3 has 175 billion parameters, which reveals the greatness of learning capacity of GPT-3. This also proves that how a crucial breakthrough in AI technology. The new technology is aimed to be used in many different industries such as software technologies and design.
Has an ability to write poetry, translate texts, establish a dialogue and even give an interview!
GPT-3 is a language predictor using the natural language processing (NLP) system. It is able to predict what should come after the content and have an ability to implement commands. This product benefits from posts and comments published by people on social media and other forms as well as encyclopaedic information on web. The algorithm uses all these texts on the Internet to generate a plausible answer and since it is nourished a lot of data. Therefore, the predictions are generally highly accurate.
As we mentioned, high quality of the texts written by GPT-3 differs it from its old versions. In fact, it is impossible for us to understand whether the text is written by human or machine. Let’s look at the capabilities of this product in general:
A Machine cannot achieve this!
GPT-3 is one of the most interesting and important artificial intelligence systems ever produced. It is easily possible to reach encyclopaedic information with a standard search engine. However, this product provides more than that. For example, with this product is able to;
• writes computer codes,
• solves language and syntax puzzles with the help of just a few examples,
• writes blog posts and academic articles,
• composes music,
• even provides an interview with historical figures. For example,
OpenZeppelin’s co-founder and former CTO Manuel Araoz interviewed with Einstein to demonstrate the capacity of GPT-3. It sounds so futuristic but do not worry, it cannot travel in time yet 🙂 However, with the data it obtains, it is possible for GPT-3 to imitate a well-known public figures.
This technology seems inconceivable and it is in our lives now. It looks like GPT-3 is the closest AI product to pass Turing Test so far.
But a little scary…
On the other hand, with these features, GPT – 3 may be assumed a little bit scary. It may be disturbing for the ones who ponder the question of “Could artificial intelligence be the end of humanity?” As a matter of fact, researchers are also concerned about the possibility of misuse of this technology. The main reason for this anxiety arises from the articles written by GPT-3 are quite similar to the ones written by humans. For example, some researches think that it may be possible to create falsified news and contents or publish inaccurate academic publications by means of this technology.
Another claim, as GPT-3 scans not only encyclopaedic information on the internet, but also social media and form contents, it was determined how GPT-3 texts are affected from racist and similar discourses. For example, it has been observed that words such as violence, terrorism and terror are used together with Islam more than other religions. Obviously, at this point, we need to look at ourselves as a human race. We have fed Web 2.0 technology with racist and sexist arguments. Therefore, the discourse of GPT-3, which is fed by these contents, is actually our reflection. This result is also like a preview of how we will influencde the next generation. After all, they will grow up with what they hear from us.
Despite all these opinions, it is a fact that there is a technological transformation that will influence all sectors positively if it is used efficiently. Instead of fearing artificial intelligence and producing conspiracy theories, we have to make the most efficient use of this technology.