ChatGPT, the incredible conversational AI, fascinates the Internet and causes a lot of stir. Everyone tries to see the possibilities it opens up. Humorous application, voice assistant, but also journalist or therapeutic aid, the horizon is vast!
“She came, she doesn’t like Macron, it’s CHATCGT“ ! This is how Vincent Flibustier, creator of the parody site Nordpress, presented last week ChatCGT, a far left-oriented militant version of the chatbot ChatGPT. The principle is the same as for his model: all you have to do is ask him a question for the AI to answer it… in an extremely clear-cut way. Pension reform, National Education, Emmanuel Macron, but also questions that are far removed from politics, such as the interest of Mondays or cats… She has the answer to everything – and, let’s face it, the result is hilarious! We in turn asked her about issues that are driving public debate at the moment, such as pension reform and the ban on hunting.
ChatCGT: the Marxist version of the conversational robot
On this last point, ChatCGT believes that “Business has long profited from the exploitation of wild animals for their own interests. Hunting is a cruel and unnecessary practice that endangers entire species and harms the balance of the environment. It is time that employers take their responsibilities and prohibit hunting.” Then, after a few requests, the AI simply decided to go on strike. We leave you test it your turn ! But, in addition to this humorous parenthesis, ChatGPT, the conversational artificial intelligence that responds exhaustively and naturally to any question – although it is sometimes wrong – offers new perspectives, and many researchers are exploring its different uses. possible, in particular through more serious applications and variations than ChatCGT.
ChatGPT: applications created with AI
ChatGPT is able to type in seconds a text that a human would take hours to write, to answer complex questions, to code, to solve equations, and even to engage in debates, all in a natural (see our article). Cover letter, DIY tutorials, press articles, series scripts… Internet users are constantly finding new ways to use it. It is therefore hardly surprising that some developers have already taken advantage of AI to develop different applications and make the most of it. Thus, there are already many tools, such as a desktop application for Windows, macOS and Linux, as well as an extension to chat live with the AI.
Another app, dubbed Playlist AI and available on iOS, uses ChatGPT to create playlists in Spotify and Apple Music from just a few words. The user should describe the style of music they want to listen to, such as “popular artists in the 1990s” Where “artists who have played a Coachella music festival”. The app then queries ChatGPT, which provides it with tracks, then generates a playlist and adds it to the music streaming account. To Originally, the application was called LineupSupply and offered to scan concert posters in order to automatically recognize the names of artists and create playlists, but also to generate them from a TikTok or the history Spotify or Apple Music – functions that are still offered.
GPT-3: a more efficient language model than Siri
Other developers have gone further by taking over GPT-3, the language model on which ChatGPT is based. This is the case of Mate Marschalko, who gives us a first glimpse of what voice assistants could be if they were based on such technology. In a video posted on Reddit and YouTube as well as in an article on his bloghe shows us the result of combining GPT-3 with Apple’s “Shortcuts” application to connect its dozen lamps, thermostats, ventilation system and cameras into a single ecosystem.
Remember that voice assistants like Siri, Google Assistant and Alexa only respond to fairly basic commands, such as “turn on the kitchen light”, “turn the heating down to 19 degrees” – energy sobriety requires – or even “close the shutters”. For more complex requests, you will have to go back. For its part, the helper created from GPT-3 is able to correctly interpret complete queries, such as “My wife is coming in 15 minutes. Turn on the lights for her outside when she pulls up“. To which he replies: “The lights should come on the moment your guest arrives.” Even more impressive, he is even able to call on his knowledge to make a decision! For example, for the order “Adjust the heat in the bedroom to a temperature you think would help me sleep better.“, the assistant retorts: “The bedroom thermostat was set at 19 degrees. Enjoy your sleep! ” Another example : “Yesterday it took me a while to fall asleep. Do you have any suggestions to help me sleep better?” It then suggests that the user take a nice hot bath, read a book and avoid caffeine.
To achieve such a result, Mate Marschalko had no line of code to write. He used the GPT-3 API to ask him to interpret conversational English and then respond in a structured data format (JSON) that Apple’s application will then execute. Except that, and this is the problem, it’s not free, because each request to GPT-3 costs $0.014. Suffice to say that the bill at the end of the month may be salty for daily use! And you have to test the result more thoroughly to make sure the AI isn’t saying something objectionable. But this demonstration remains an interesting experience that could perhaps give ideas to Apple, Google and Amazon, who knows!
ChatGPT: is AI preparing the future?
The use of an AI like ChatGPT could bring about profound changes in many sectors and professions. It could, for example, profoundly change journalism. Thus, the site specializing in new technologies CNET has engaged in a small “experiment”. As he explains in his article, he let the AI write about 75 articles before having them proofread, checked and edited by a member of the editorial staff. The papers were signed “CNET Money”, a rather opaque name to which have since been added a mention indicating that the writing of the subjects had been assisted by an AI engine as well as the name of the journalist who validated them. This test was conducted to see if artificial intelligence could replace an editor on a purely informative production, without much added value, with the aim of optimizing the time of journalists so that they can devote themselves to more complex subjects and of better quality, but also to see if it was possible to use it to optimize the readability of the article, its referencing or its structure. However, this raises ethical and transparency issues…
The field of medicine is also very interested in the technology of OpenAI. Indeed, researchers at Drexel University in Pennsylvania published a study revealing that GPT-3 would be able to identify certain clues in a person’s writings which make it possible to identify the first stages of Alzheimer’s disease, which affects no less than 55 million human beings in the world, with a 80% efficiency rate. More precisely, it identifies certain characteristics of language that make it possible to predict dementia, and therefore to detect the disease at its early stage.
Others are also interested in a possible use of ChatGPT in the field of psychology. The non-profit startup Koko, which offers psychological support via online chat, explains on Twitter having conducted an experiment on 4,000 people, or about 30,000 messages. She provided people who contacted her with messages written by the AI – they were still supervised by a human being – which were rated higher than those sent by a real person and reduced the response time by 50% . On the other hand, when users learned the truth, they simply stopped trusting the service. Still, this questions the capacity for empathy – or at least simulation – of chatbots. An experience that echoes that of a young woman who used GPT-3 to “dialogue” with a younger version of herself. Could AI be useful for mental health?
i kept diaries for about 10+ years of my life, writing almost everyday about my dreams, fears, secrets
the content ranged from complaining about homework, to giddiness i felt from talking to my crush
some days were very mundane, some rather insightful pic.twitter.com/CzA1C20U4a
— michelle huang (@michellehuang42) November 27, 2022
Code, 3D animation, writing, journalism, medical and even legal… AI seems to open up a myriad of possibilities! In any case, several digital giants see in ChatGPT and other OpenAI technologies something to revolutionize the future. This is particularly the case of Microsoft, which has invested billions of dollars in the company in order to obtain an exclusive license for GPT-3 and thus integrate it into all its services, such as the Microsoft 365 office suite and its engine. to search for Bing. Enough to panic Google, which has launched a “red code” and reorganized various departments to move forward on its AI projects, including Sparrow, which could well be a serious competitor to OpenAI’s chatbot. But all is not rosy, far from it. You only have to look at hacking to see this, as hackers are already using artificial intelligence to create new malware and to sell fraudulent apps.