On Monday, 13 May 2024, the world awoke to a new announcement and one that I feel is going to have a marked influence on the way humans and artificial intelligences interact. No longer are we relegated to text-based questions and answers. In fact, human-like conversations between humans and AIs have now officially entered the realm of mainstream interaction.
Soon after the announcement I had the opportunity to test out openAI’s newly released GPT-4o model (the ‘o’ standing for ‘omni’), and I was blown away with its (or should I say her?) capabilities.
As a high-level software programmer and educator, and as somebody who is genuinely fascinated in AI and has been since my many years at university, I have a vested interest in keeping up to date with the technology. This affords me the ability to impart newly acquired knowledge onto my students and to implement this knowledge into my own projects. I can say that GPT-4o is a major step up when it comes to conversational enquiries and to query responses.
I started by asking the older GPT3.5 model to generate a Java class that I could use to interact with a data file in JSON format, an open and lightweight language-independent format for storing and transporting data in a readable text-based structure. The AI model responded with perfect code using the ubiquitous ‘json.simple’ library that I could copy and paste into a project in my NetBeans IDE, my IDE of choice for Java development. Not only did she produce the code listing, but also a text-based explanation on how the code works and what needs to be done to modify it for one’s own project. With very little modification, the code worked perfectly.
To test the AI’s capabilities, I then asked if she could repeat this request but use the Jackson JSON library instead. Five seconds later, I had a code listing that had been completely rewritten to incorporate this new library. Once again, upon copying and pasting it into a new project, it worked brilliantly without any major changes.
My last test for GPT3.5 was to ask her to rewrite the code but this time in Python. Again, without any fanfare, a Python code listing was produced, which worked perfectly when copied and pasted into a new project in PyCharm, my default Python IDE.
Two things struck me: firstly, the speed at which my requests were answered, and secondly, the amazingly wide capabilities that this AI model possesses.
I then turned my attention to the newly announced GPT-4o model, which now no longer needs to be prompted using text, but rather, voice and video is now the order of the day. I printed out my modified Java code listing from my first test above, and then asked GPT-4o to look at my code and to please tell me what its function is. She proceeded to tell me, in quite a lot of detail, not only what the code does, but also how it does it. Dare I say that her ability to reason seemed uncannily human? The inflections in her voice coupled with pauses in the correct places would, had my eyes been closed, have me believe that I was talking to a human being.
One disconcerting fact was that the interaction took on a slightly flirtatious tone when further questions were asked. As a member of the so-called Gen-X generation, I do not need my AI to flirt with me. I simply need answers. I suppose though that these new AI models are not catering to my generation. After all, it is the younger generations that are going to be sharing their world with them in the future.
We sure are moving rapidly into a brave new world!
| Tel: | +27 11 543 5800 |
| Email: | malckey@technews.co.za |
| www: | www.technews.co.za |
| Articles: | More information and articles about Technews Publishing |
© Technews Publishing (Pty) Ltd | All Rights Reserved
printer friendly version