At Genbeta we have already seen what GPT-3 is capable of, the OpenAI language model that is willing to change things in the field of artificial intelligence. We even told him to write us a full article, with surprising results. After the publication of this language model, many have tested its capabilities. The last of these examples has been that of Lucas Rizzotto, better known as Lucas Builds The Future on YouTube, who has given 'life' to his microwave through GPT-3.
With the help of this language model, this content creator has trained the AI with a document of about 100 pages, giving it the necessary context and explaining to this artificial intelligence that the microwave is actually an imaginary friend that Rizzotto had in his childhood. The story however took a totally unexpected turn, as the microwave was 'reborn' with unfriendly intentions.
Magnetron is born, the veteran microwave in the First World War
The name he gave him when he was just a child was Magnetron, and according to him, he was an 'English gentleman' born in the early 20th century and a veteran of the First World War. Of course, this was all in Rizzotto's imagination, but he thought it would be fun to give the AI this context and then insert all this information into an Alexa-enabled microwave he was able to purchase recently. Nothing more and nothing less than a writing of about 100 pages detailing the memories of his imaginary friend.
Lucas managed to insert all this information into the smart microwave by modifying it and adding a microphone and speakers to communicate with it. To do this, it made use of GPT-3, incorporating this language model in the microwave and thus taking on 'life'. Rizzotto inserted GPT-3 with the API of the microwave itself, so in addition to conversing, it could also carry out the typical functions of this appliance without problems.
According to Rizzotto, talking to this microwave was equal parts beautiful and mysterious. It really did feel like talking to an old friend, and while not every interaction was perfect, the experience held up well enough to be believed. In addition, by giving him context about things that he only told his imaginary friend, he was the only one who knew them, bringing them into the conversation in an organic way, according to Rizzotto.
This imaginary friend began to take on a darker and darker tone
However, in the conversations, Lucas began to detect an increasingly aggressive tone, having no idea what was happening. In fact, once the microwave asked Rizzotto to enter it. She played along, telling him that she was already inside him, and once that was done, the microwave decided to turn on. That's right, Magnetron wanted to kill him by keeping it inside him.
Rizzotto didn't know why, so he asked him directly. The microwave replied that 'it wanted to hurt him just like he did' a long time ago. At this moment, Lucas discovered what was happening to him, and it was that he had not spoken to his imaginary friend for about 20 years, making this AI 'understand' that he had abandoned him. As much as he apologized, Magnetron did not relent, and Rizzotto thought that he should turn it off one last time.
The point that Rizzotto wanted to reach with this is that, it does not matter that this artificial intelligence was not objectively 'real', given the behavior it acquired and the conversations it was able to have with it, for him it was, and it was what the only one that mattered. Although the intention with this article was simply to show you the capabilities of this powerful language model, we have ended up discussing the 'humanity' that an artificial intelligence can acquire.