Could the new artificial intelligence that caused a stir this week replace Google advantageously?
It must be said that she does not just provide you with a list of links to consult without guarantee of results when you ask her a question!
ChatGPT, Google (much) better?
Could ChatGPT be the future of internet search? We can ask ourselves the question after the results he produced following the various questions he was asked this week! As a reminder, ChatGPT is a chatbox endowed with an artificial intelligence based on the GPT-3.5 language model and developed by the startup OpenIA.
Throughout the week, many Internet users marveled at the answers provided by this AI, which was able to offer stunning and detailed results when complex questions were submitted to it. Its ability to solve very advanced math problems, for example, has led to many jokes about the future uselessness of homework, as every student may one day have an AI ready to do the work.
But more generally, if the technology of this chatbox improves, and the data flows into it in large numbers, it could eventually make a search engine like Google obsolete. Take the example of a movie buff looking to watch the first film directed by Akira Kurosawa (that’s Uma , for culture). While he could waste time looking for an opportunity to see the film in question by browsing on Google, ChatGPT could tell him directly which streaming platform to go to, or failing that, a site offering the work for rental.
But a dangerous Google?
But while time savings and superior response quality seem clearly assured in this scenario, another problem arises. Couldn’t ChatGPT’s precise explanations be used for harmful purposes?
We obviously think of individuals wishing to manufacture explosives for example, and who would use the technical knowledge of the Chatbox to quickly learn how to build small harmful devices. A difficulty that OpenAI claims to be aware of and on which it is already working. ChatGPT having already been configured not to show the same kind of political or ideological biases spotted on previous artificial intelligences, we can hope that the start-up will once again succeed in reducing the risks.
And for those who still haven’t tested ChatGPT, go to this link .