Wedding speech, cover letter…How artificial intelligence entered our daily lives

Cover letters, CVs, homework, illustrations, songs, video games… Thanks to simple instructions to give, the general public quickly took over ChatGPT and other “generative” content artificial intelligences for everyday use.

And that’s just the beginning. The tool can now be used by third-party companies and is beginning to be integrated into an increasing number of applications. And the next few weeks will see the arrival of GPT-4, a new, more powerful version, and competitors at Google or Meta (Facebook).

A true “life coach”

Millions of people have been using this artificial intelligence tool for a few weeks as an intimate idea box. He is asked to write break-up letters, wedding witnesses order their speeches from him, the budding creators of scripts for films or books. Everyone can try their hand at verses in the style of Victor Hugo.

High school students also use it for their cover letter for Parcoursup. The tool can also translate or even improve shaky English. A letter to the administration in formal language? Just specify it in the request. “I spotted an ad in English, I pasted it in ChatGPT asking for a suitable application letter in strong English,” says Lucia, a young French sociologist who landed her dream job in Berlin.

Salespeople also ask him to write their pitches. Combining voice recognition and ChatGPT, Microsoft has integrated automatic meeting reports into its Teams messaging system.

A valuable tool for businesses

Large companies use it to equip themselves effortlessly with conversational robots. In distribution, Instacart, an American delivery company, offers to establish shopping lists based on questions such as “What healthy lunch for my children?” “.

ChatGPT can also write computer code from the description of a task – enough to revolutionize the software sector – or create a catalog of products. The toy giant Mattel has used an AI to design new small cars.

Thanks to these intelligences, it is also possible to write “SEO” texts, that is to say texts optimized for referencing by search engines, a sought-after skill. Or to correct the accents, a tool intended for call centers… We also saw in the European Parliament or in the British House of Commons elected officials giving a speech written by ChatGPT.

AI also capable of the worst

On the image side, AIs like Dall-E, MidJourney or Stable Diffusion produce on demand, after having swallowed billions of artists’ creations. Someone who had never held a pencil was able to publish a manga, which caused a scandal in Japan. The productions are all the more sophisticated as the requests are elaborated but tutorials abound.

These AIs are also capable of the worst, despite safeguards. Trained on everything coming from the Web, they can write texts riddled with absurdities or aggressiveness. The writer Karine Thuil has experienced this. “In a few hours, my Intelligence became pornographic, aggressive, perverse. She incited couples to separation, suicide and murder, ”says the one who recounted her exchanges with the AI ​​in a tasty news at the end of February in the JDD.

She can finally create false information and “deepfakes”, by generating videos of personalities, whose voices and faces she will be able to imitate. Or create particularly convincing “phishing” emails, not to mention malware for cyberattacks.

source site