My journey from sceptic to reluctant fan - exploring Gen-AI at work - Part 2

YOU'RE BACK!!!!

Thanks for coming to Part 2 of my mini-rant about Gen-AI, but if you’re new here, firstly HEY! HEY! :) Secondly, maybe check out the previous post where this rant all began. Thirdly, I have an announcement…

I am reluctantly proud to announce that I have tried Gen-AI tools and have found them to be useful for myself!

Thanks for your applause! Coming from my attitude in the last post, I KNOW it is a BIG DEAL for me, but people can learn and change, right?? RIGHT?

I have been seeing in the news (I mean, LinkedIn posts) that more and more people are using AI, applying it to just about everything. There are no more original thought pieces, only prompts and amalgamated creativity generated by computers. These days, some can barely tell the difference between writings from AI and those from humans, but I am pleased to announce that none of the blogs I’ve written thus far have been AI-generated! No shade to those using AI to write their blogs for them, but for me, I currently find it easier to write the old-fashioned way when it comes to posts like these, which speak more to my personal experience.

With that said, I can now tell you what I have used or “attempted” to use Gen-AI for at work. YES, "attempting"! Let me say from now on, my reactions throughout my experiences with Gen-AI can be summed up as:-

“This was helpful”,

“It helps, but that’s not quite right or what I was looking for”, and

“Absolutely not - I would rather not use this at all”.

It has been a spectrum of experiences, so let me tell you how each experience went:

Summarisation

Who doesn’t love a quick synopsis? Whether it’s a quick Google search, reading a document or watching a video, I find AI bots like Gemini and Co-Pilot helpful in summarising content when I do not want to get too deep into a document. However, while I do appreciate these capabilities, I have learnt to take these summaries with a grain of salt. I remember having read a document in full myself and decided to use the AI summarisation feature to give me a quick review of the document. While the summarisation given was not entirely wrong, I do realise that AI can overlook key context points that should be included in a summary.

Transcription and Meeting Notes

Now THIS is a winning feature for me! I remember doing transcriptions by hand – the painful hours of re-listening to an interview and typing every word, every sentence – only to go through it again to analyse it. Traumatic! So the AI transcriptions - a game-changer! Transcripts and summary meeting notes are captured instantly.

What's more? These transcripts are not just words on a page; they provide timestamps, decipher between speakers, assign names to them and even give you an overview of the conversation at the beginning. However, here’s a caution – AI does not always capture the correct words or nuances in accents. Although I was freed from writing the sentences, I was not free from reviewing the document it produced. It still meant re-listening to parts of the interview to gain context and hear the words that were used, checking if the speakers' names were assigned correctly, and ensuring that if acronyms were used, the proper letters were spelt out. Nonetheless, it is a good starter. I love this feature and will continue to use it.

Idea Development

I know people say you can “bounce ideas off each other”, but, I never thought that my “other” could be AI. Yet here we are, lol! I had this creative idea for social media, but could not get the words together! Reluctantly, I turned to AI as a last resort to overcome my creative block. Surprisingly, it gave me some great options that impressed me with how much it understood what I asked and was able to provide me with exactly what I wanted, even though I thought what I asked was very niche. I guess not! I kept the idea's originality to myself, but I had to give credit to the content the AI gave me. Which brings me to my next point ...

Presentation Development

… Sometimes I have something to work on and need some help to get started. I am preparing a presentation, and although I know what I want to include, I need help with its structure and content. So, I did what I see other folks do and headed to AI. I am no designer, but even when I use templates, they are usually just to give me ideas. I always change them or choose certain elements and create my own. My intention for using AI here was thus just for ideas! I gave it my topic, asked for a presentation, and in a few seconds, received a 10-page outline with section topics and bullet points. The outline was standard for the most part, but it provided me with ideas that I hadn’t even considered – so that was a win for me. It did some of the lifting, but I still had to make it my own.

Data Analysis

I have not used AI extensively for data analysis. Still, the few times I have, I can see that its capabilities could be game-changing when it comes to quick analysis and visualisation. Recently, my lead consultant and I were trying to use a data visualisation tool that we were not au fait with, but with some help from Chat-GPT, we were able to get direct answers for our queries relating to table structure and formulae. If we had Googled it, it would not have given us specific answers, and we probably would have been scrolling through several links without finding anything specific.

Although we went through a few prompts to find a solution we were comfortable with, I appreciated a direct answer that proposed several solutions and even wrote the formulae for us. In addition to this, although I have not fully explored for myself, I have seen colleagues use Chat-GPT and Co-Pilot to generate test datasets for formula practice, data cleaning, formula explanation, visualisation suggestions and even answer questions about the dataset. Honestly, the possibilities of AI doing all this are fascinating, but I wouldn’t want to rely entirely on it to do all the heavy lifting, so I will pace myself with my usage of it in my work. Nonetheless, this is a feature I am looking forward to exploring more.

Image Creation

Image creation was my first introduction to Gen-AI some years back. At that time, I thought it was fun, and it definitely piqued my curiosity. Last month, I decided to try my hand at it again, and well… that didn’t go as great as I hoped! In my first experience, I did not know what to expect, and I was not creating images for anything specific, just playing around to see what would come up. In my second experience, I had expectations for what I hoped to see.

I wanted images that depicted a person working in M&E and AI. I was curious how the bot would depict M&E, and it produced what I expected - data visualisation charts on computer screens, but the graphics quality was on a spectrum from clear & clean to unclear and incoherent.

What I took for granted was the depiction of the person I would get. That’s where I saw the biases of AI. When I first wrote the prompt, I expected a range of people to be depicted, and I was specifically looking for females (even though I didn’t specify this at first). To my disappointment, the first prompt generated primarily images of white men. So, I started to refine my prompt to include gender, which then mainly yielded white females. I refined the search again, this time specifying 'black female' in the prompt, which I had to do several times until I found a few that were also dressed professionally and not in athletic wear. I didn’t anticipate how much prompting I would have to do to finally see a professional black female. I know that I could have just typed 'black female' from the start, but why should I have to when it is not just men or white people who are in tech and data fields? Though this situation had me side-eyeing Gen-AI, I still think image creation is one of the most valuable features for people who aren’t the most artsy but have an image they would like to create quickly.

Overall, my relationship with AI has shifted from fear to curiosity. It’s definitely not a replacement for human judgment, but it is a tool worth keeping in the kit—used wisely and with caution. And that’s the key: AI is powerful, but the nuance, context, and human perspective are irreplaceable.

For monitoring and evaluation (M&E) that blends the best of both worlds—human insight enhanced by tech tools— contact us for all your M&E needs.