My Journey with AI

 Artificial Intelligence (AI) has made significant strides since its inception, with advancements in large language models revolutionizing its accessibility, but AI has a long history. I can still remember the Jeopardy episodes between Watson and Ken Jennings and Brad Rutter (video summary). I worked at IBM at the time and after the tournament, I started seeing some new things coming out in AI. My first direct interaction came when I was asked to participate in an experiment in sentiment analysis. I provided access to my Twitter feed and was provided a report on the sentiment of my feed that was really pretty accurate. 

The first real AI project I did was a bit of an experiment. I was very involved with the Project Management Institute (PMI) at the time and came up with an idea. I connected with a couple of people in the Watson group to try it out. The hypothesis I came up with was "Watson can be used to identify project risks in a statement of work." This was before the days of large language models, so we had to train the tool to understand what project risk meant. I remember tracking down a PDF version of a project risk book and using that, plus a set of questions and answers, as training material. We then ran a statement of work through, and Watson was able to identify some basic risks. The results of our experiment were presented at the 2016 PMI Global Conference. 

The introduction of large language models, including ChatGPT, has transformed the AI landscape, simplifying the user experience. Users no longer need expertise in training models; instead, a basic understanding of prompt engineering suffices to obtain desired information. However, a crucial challenge arises due to the knowledge gap between the average user and the model's training process.

While ChatGPT and similar AI tools offer impressive capabilities, blindly accepting their responses as accurate would be misguided. Verifying the facts and information presented by AI tools becomes crucial. Recognizing this need, regulatory efforts, such as those by the European Union, are underway to establish guidelines governing the use of AI.

The evolution of AI from early experiments to the era of large language models has revolutionized its accessibility. However, as AI tools like ChatGPT become more prevalent, it is vital to exercise caution and verify the information they provide. By understanding the journey of AI and embracing responsible practices, we can harness the power of these tools while ensuring accuracy and reliability in an evolving AI landscape governed by emerging regulations.