My Experience with ChatGPT so far
AI / Machine Learning in general and ChatGPT in particular currently experience a big hype and I actually do share that there’s something ground-breaking to it. I even believe that consulting an AI-driven question-answering system will replace the classical searching on the Internet on a large scale.
I bet there‘ll soon be a new word for doing exactly this. Just like to ‚google‘ something, there’ll be to ‚chatgpt‘ something or something like that. Needs to be more concise, though, easier and shorter to pronounce. Maybe to ‚ai‘ it or so.
So when do I find it personally useful to use ChatGPT? I would summarise it as follows so far:
- Help with programming issues (not really surprising, but helpful indeed)
- Ask general knowledge stuff, even medical advice. In particular, I like how it is providing a well structured, systematic presentation on a topic. But you will have to ask explicitly for pointers for further reading, and then those are often seem not publicly accessible, unfortunately…
- Creativity: ChatGPT is actually good for writing articles and stories, as well as creating images. The only caveat is, that you still feel, on a higher level, the human, emotional touch missing. And you sometimes need to specify things that among human beings would go without saying. Such as when people talk to each other they normally also look at each other (in the case of drawing an image). Nevertheless, I find the drawing functionality sometimes really helpful for creating key visuals for my texts, e.g. for my blog posts. Just like the one used here 🙂
- Translation of texts – does take context into account, which is indispensable for a good translation.
- Upload a photograph and ask what is on it: ChatGPT does a pretty good job with image recognition, as far as I can see. Of course, you need to have a decent picture, but then you can get help finding the right name for something. Having the right name again is key to finding further information on it …
Some things I like when using ChatGPT:
Fault tolerance in my input to ChatGPT: After submission, I often find that I misspelled something or swiped a totally different word than I intended to on my mobile phone. ChatGPT usually still understands what I wanted to say. On a higher level than a good search engine would, I shall think.
Code switching / Change of language: I can talk to ChatGPT in any language that I currently find easier to express myself in and will be just as well understood. ChatGPT usually also answers in that same language. That is not the case in a classical search engine such as Google. There, I am more or less forced to stay in my language and region, get directed to translated websites, which I would prefer to read in the language I have expressed my search terms in – if shown at all, if e.g. relevance scores are low because of my current locationd.
Leave a Reply