OpenAI’s spring news has left artificial intelligence enthusiasts with their mouths open, with a new model that allows voice interlocution and that allows visual recognition through the devices’ cameras, among other new functions.
To present the new product, GPT-4o, Sam Altman’s company has published a series of videos on Youtube in which members of your team chat with the artificial intelligence in different ways to show off the model’s capabilities. Precisely, it is in the way in which OpenAI employees have spoken with the chatbot where some users have noticed something somewhat worrying for the future.
And it may be a small thing, but taking into account the impact that new technologies have on our behavior in the long run, it is not unreasonable to think that details like this can greatly influence users. Those who have seen several videos or fragments of the OpenAI presentation will have noticed: the company’s team does not stop interrupting GPT-4o when talking to her.
Perhaps because the response offered by the language model is somewhat long, or because the AI is not completely following the line that the team seeks to teach, there have been many times in which those from the brand have spoken over GPT while it is was still talking.
Users like @Mappletons have not missed this nuance: “I imagine that the least interesting use of GPT-4o is to emulate an uncomfortably flirtatious and slightly condescending woman who you have to actively interrupt if you want to say a word,” she points out. User @wcandillon adds: “I found the way they interrupt the AI in all the demos a little disturbing. “It sets a horrible example for conversations between humans,” she comments on X.
Maggie Appleton
@Mappletons
I imagine the least interesting use of GPT-4o is emulating an uncomfortably flirtatious, mildly condescending woman who you have to actively interrupt if you want to get a word in edgeways. https://t.co/vG3BAEo8iN
May 15, 2024 • 15:02
691
9
The truth is that, as others have commented, talking over the AI even if it has not finished giving its response can serve the OpenAI team to demonstrate the ability of its model to maintain a conversation, even with the added difficulty. of an interruption. Likewise, it can bring realism to the use of technology, since cuts or overlaps between voices are common in human conversations.
But of course, there will be no shortage of people who give an extra layer of meaning to this circumstance: “it teaches people to be rude in the long term”, “I’m afraid that children will learn disruptive behaviors”, “their interruptions of the model were somewhat provocative”, “it is not a conversation between equals, it is more similar to how someone talks to their servant” are some of the comments in this regard that can be found on networks. The main fear would be that, due to the ease of interrupting the AI at any time, people could adopt this bad habit and interrupt other people on a daily basis.
OpenAI is not immune to criticism
Although the news announced by OpenAI places the company at the forefront of technologies of this type, there was no lack of criticism of the product shown. The parallels between the new model and the one starring in the dystopian film ‘Her’, as well as the sympathy and affectionate comments offered by the AI, have been criticized for several reasons. Among them, they accuse OpenAI of taking advantage of gender inequality (although there is also a male voice available), or of trying to create a model that establishes personal relationships with users to generate a relationship of dependency. This may all be nonsense, or it may not.
