It's easier to stumble when you run desperately. The launch of ChatGPT a little over a year ago caught so many companies at a different pace that the race for generative artificial intelligence within reach of the general public has become a fierce competition between them to demonstrate that they have not been left behind, but quite the opposite.
This haste is what Google appears to be, whose work in the field of AI is much more important and relevant than its missteps show when it announces them. After presenting last week its new great language model, Gemini, capable of surpassing ChatGPT's GPT-4, the company led by Sundar Pichai ended up admitting that it made a montage of the presentation video so that it showed not what it does. now its AI, but what it can do. Also at a gallop, something unusual in the processing of legislation in Europe, the EU has reached an agreement for an AI law, the first in the world, although the different states have ended up imposing on the European Parliament some safeguards for the use of this technology in facial recognition in public spaces, one of the aspects that raises the most suspicions on the part of human rights defense organizations.
It is the second time that Google screws up with an AI. It did so by presenting its Bard chatbot—it doesn't work in Europe—by presenting an erroneous answer about the James Webb telescope, and it does so now, but the search engine company has promising material beyond wanting to put the cart before the horse. in their presentations. Google has created three versions of Gemini, a smaller and more limited one, to be installed on mobile phones, called Nano; another more powerful one, called Pro and that works through an internet connection, which would be equivalent to the first GPT-3.5 from ChatGPT; and its most capable model, called Ultra, which will presumably be the one it would offer to anyone willing to pay to use it. Using this technology requires enormous computing power, therefore energy. Everything is paid. The first sample of Gemini Ultra showed an AI capable of interpreting in real time what it saw on video. Let's imagine that instead of a recording you were shown a live camera. Science fiction robots are getting closer to us every day.
That idea is already on social networks, which have led us to consume more and more content to our liking, always according to our criteria. By instinct we do not follow those who disagree with our ideas, only those who reaffirm us, and thus we shape the world in our image and likeness. This behavior can be accentuated even more with artificial intelligence. Tailor-made couples arrive. In the film Her (Spike Jonze, 2013), a human falls in love with an AI, Samantha, with whom he becomes intimate, even having a kind of virtual sexual relationship. For those who have not seen it, I will not reveal the ending, but building a tailored relationship, far from the singularities that each person has, does not have to lead to happiness. With tailor-made AI-pairs there is nothing to disagree with. It's like loving yourself. The myth of Narcissus is more current than ever.
