Today, the start-up claims more than 100 million weekly users of its conversational AI service in 185 countries and does not intend to stop there. It is now possible to use ChatGPT without creating an account. The measure is being rolled out gradually, “with the aim of making AI accessible to anyone curious about its capabilities.”
If the stated objective is, in appearance, completely disinterested, it is much less so in reality. The startup plans to use information users provide to ChatGPT to improve its models. Enabled by default, the function can however be deactivated for those who are against the idea.
According to OpenAI, an account offers many advantages
As a reminder, creating an account to access the ChatGPT service offers, among other things, the possibility of saving and viewing the history of your conversations, sharing them and unlocking additional features such as voice conversations and instructions. personalized. It remains to be seen whether this argument will work in favor of creating an account.
In addition, OpenAI specifies that it has implemented additional content protection measures for this experiment, such as blocking text commands and text generations in a greater number of categories. And this point is very important.
What about this restrictive policy?
Our colleagues from TechCrunch asked OpenAI which categories are affected by this policy. The company responded as follows: “The logout experience will benefit from existing security mitigations already built into the model, such as opting out of generating harmful content. In addition to these existing mitigations, we are also implementing additional, specially designed safeguards to process other forms of content that may be inappropriate for a offline experience.
And to add: “We examined potential ways in which a disconnected service could be used inappropriately, drawing on our understanding of GPT-3.5’s capabilities and the risk assessments we have conducted.” Remember that if the start-up is committed alongside other technological players to promote a certain transparency relating to online content, the fact remains that, last January, it opened the door to applications considered divisive to say the least.
The start-up has thus modified the policy for the use of its models, erasing the terms “military and war” from the list of prohibited uses. And therefore opening the door to military applications of its technologies.
Do you want to stay up to date on the latest news in the artificial intelligence sector? Register for free to the IA Insider newsletter.
Selected for you