June 7, 2024
It's been over a year since we first heard about the revolution that artificial intelligence was expected to bring to practically every existing sector. However, for most of us in our daily work, AI hasn't disrupted the balance as much as expected. Of course, this depends on the sector and the type of company. In any case, I'm pretty sure that in most offices, besides using ChatGPT for fact-checking (a highly risky operation that I discourage due to the fact that these models can easily reproduce falsehoods), correcting some emails, or writing articles, the full potential of available technologies is not being fully exploited. With this introduction, I didn't mean to be critical of the progress made by AI, much less critical of the potential of the technology itself. In fact, I am among the curious and supporters of the theory that this technology in the medium term will perhaps represent the greatest cultural revolution ever. To contribute in my own small way to this revolution, I wanted to create a personal assistant entirely made with artificial intelligence.
ChatGPT has been an incredible revolution, capable of showcasing to people the true potential of AI that had previously remained hidden. It made it possible to interact with software without the need to know a programming language. From a technological standpoint, ChatGPT represents a fascinating convergence of physics and computer science, but its true revolution, in my opinion, has been to communicate the potential of AI to the masses.
However, if we ask why ChatGPT hasn't revolutionized our way of working, the answer is rather simple. If we were to personify ChatGPT, we could consider it as a kind, helpful, and knowledgeable colleague, capable of providing a wide range of information. However, we cannot delegate any tasks to it. We can ask it to correct an email, but not to send it for us. We can't even ask it if we have unread emails or to schedule events in the calendar.
That's why I decided to develop a Personal Assistant based on ChatGPT, capable of performing these tasks and facilitating the revolution that is being talked about so much. Despite it no longer being the most advanced generative AI model in the world, I chose ChatGPT as a personal assistant because of the unbeatable tools developed by OpenAI for us developers.
In this article, we will explore a version of ChatGPT hosted within Google Sheets, although this detail is not relevant. What really matters is that, compared to the traditional version of ChatGPT, thanks to a service from OpenAI, this version is able to interact not only with the data I provide to ChatGPT, but above all with the functions we can ask it to activate.
I mentioned accessing the data, and it might be easy to think that I mean training a GPT model on my data. However, this is not the case. Training a model on one's own data is a complex task and could expose our data to biases common to these tools. What I did in the example you will see shortly was to directly connect a database containing the data I want to share to ChatGPT. This way, when I ask ChatGPT to provide me with certain data, it knows to query the database directly. This approach eliminates the need for long periods of AI training and reduces the risk of bias, as the data it provides comes directly from our databases. For example, I connected a database containing information about my cats to ChatGPT.
.gif)
Yes, finally ChatGPT is able to activate any function I describe. No matter what the function is, simply give the textual command to ChatGPT and it will be able to initiate it and even send any variables to the function. This means that ChatGPT is no longer limited to providing information, but is starting to actively perform tasks on my behalf. It can now send emails, schedule events in the calendar, create invoices, add phone numbers to the address book, and much more. Essentially, I finally have a little assistant that really saves me time. For example, I ask ChatGPT to send an email on my behalf, and it asks me for all the necessary information to complete the operation.
.gif)
Today, artificial intelligence represents an incredible tool. As Francesca Rossi, Global Leader on AI Ethics at IBM, suggests in her book 'The Edge of the Future,' we must adopt the concept of 'Augmented Intelligence.' This means pairing AI tools with human figures. This synergy, in many sectors, produces outputs of superior quality and with a success rate significantly higher than complete replacement of humans with artificial intelligence. If you're interested in the topic of artificial intelligence, I invite you to listen to the episode of the Grande Giove podcast where Daniele and I chatted with Francesca Rossi.