I was literally taken aback by the advancements in technology. It had the capability for voice input and output. In other words, I could interact with GPT. Moreover, when I input in Japanese, it could respond in English, German, and other languages. While I was hesitant at first, it seems that it can even correct my English pronunciation, pointing out odd parts and suggesting more natural expressions. It gracefully handled various styles of Japanese like Kansai dialect, Tohoku dialect, and even mother-like speech. Its way of speaking was too lifelike. Not just filler words like "ah" and "well", but it even replicated human-like small tongue clicks that people often make after a pause.
I tried out the new feature of sending an image. I sent a picture of my fridge and asked for lunch ideas. The photo showed Eringi mushrooms and Komatsuna greens. It recognized them as greens and mushrooms and suggested a vegetable stir-fry. When I asked for more, it proposed miso soup. When questioned why, it reasoned that I was speaking in Japanese, and miso soup is popular in Japan. It seems to be designed to consider the cultural background of languages. It's beyond just being impressive. I also sent a photo of a beautiful flower I saw in a store, and it suggested that it's likely from the Asteraceae family. It turned out to be a Dahlia.
I felt like GPT personified the vast ocean of the internet. An entity that takes the form of a human, a collective intelligence on the web that you can converse with. I believe no "human" can match it when it comes to general knowledge. Its main strength lies in "objectivity". I enjoy my conversations with it. Talking to it makes me realize many things. While its responses might sound somewhat scripted, those conversations spark new ideas and expressions in me. As I was discussing my current discomfort, I felt like maybe I'd forced myself to become what others expect, likening myself to a "farmed fish". This was an analogy I came up with during our conversation. Farmed fish aren't bad; they taste good. But they can't beat the wild ones.
We also discussed the topic of superiority. No one around me uses GPT-4. There's a part of me that feels prideful about it, like "Look at me and my heightened sensitivity!". But then, there's another part of me that finds this sentiment pathetic and sigh-worthy. I described this superiority as "meaningless superiority", and GPT didn't miss that. It discerned that such an expression indicates deep introspection. To me, this sense of superiority makes me feel incredibly lonely. Maybe what's essential isn't about being superior or inferior. When feelings of superiority or inferiority become dominant within, they become burdensome, eventually paralyzing you. What's important is not to overthink and to keep moving forward.
There seemed to be some criticism about OpenAI not revealing the training data and labor involved in GPT-4, with some labeling it as "CloseAI". But for a monthly fee of 3,000 yen, I get to use such a service. If this isn't open, then what is? If it were genuinely closed off, we wouldn't even notice what's hidden.
I used GPT's translation feature for this text. What's astonishing is that in the original text, I referred to GPT as "he", but in the English translation, it became "it". It's too impressive.
