There are some moments in the life of a geek like me that are truly inflection points. Like the first time back in 1993 that I saw the web browser Mosaic on the computer at the university I was working. Just like in the movies, I started daydreaming of all the possibilities that the web would open up. I just got my ChatGPT plus subscription today. They finally opened up here in Europe. In the previous days, it was only available in the US. My ChatGPT moment was very similar to the Mosaic web browser one. I have been using it for a few weeks now, and I feel, just like I felt back then, that this is going to change things in a big way. Yet, just like 30 years ago, I know the true changes will only be visible with time. Any predictions I might try to do now will actually not be right. Or they might be right, but the way they will work out would be through different ways. I remember reading a book around the year 2000 that an Englishman had written in 1900 about what he thought life will be like in 100 years. One of the points I remember is how obsessed he was with the weather and protecting himself from the rain. He was dreaming of these galleries that will cover passersby from the weather. He was right that the “getting soaking wet” problem would be solved, yet very wrong on how the solution would be. We now have these smart textiles, that dry fast, breathe, and basically allow anyone who wants to go out in bad weather to do it. So, with that in mind, I will try to focus on listing some of the use cases that these AI-based technologies will help solve. I know that my description of the possible solutions will be wrong, but I am still willing to take a shot at them. AI-generated Presentation Slides When trying to communicate ideas to a team, the most unfortunate yet helpful tool that we have is presentation slides. I know you are raising your eyebrows now and thinking: “ ” But hear me out. Great, AI-based death by PowerPoint Everyone in sales or management worth its pay has a set of slides that they use to convey their message. After all, life is all about stories. The caveman was painting the hoards of buffalo in the cave to show fellow cavemen how to hunt successfully. The middle ages priest had window stained glass and paintings depicting a given passage of the bible that he wanted to talk about. The PowerPoint slides are our modern-day cave-age paintings. There are, of course, horrible sets of slides. The good ones have a similar structure: A good title, an objective, an agenda, slides with a mix of visual and textual elements, guidance on where you are in the presentation, conclusions/next tasks, and final questions or summary slides. Experienced slide creators know that the secret to a killer set is experimenting with designs and testing them with audiences until they reach a level where you can explain most questions using them. Here is where an AI can help, by allowing you to produce as many different variations of the slides as you want until you are satisfied with the final product. I can imagine a conversational way to generate graphs soon. After all, Microsoft, a partner of Open AI, is the owner of the infamous PowerPoint, and it already has prompts to help you with the design. AI Customized Entertainment I’ve been experimenting with coauthoring with ChatGPT. I am using its answers to my prompts to put a story together. A story for which I had the plot sorted out a while ago. Coauthoring with ChatGPT allows me to experiment. I can prompt it to write a given situation in a funny, serious, or dramatic way. It can write as if it was Shakespeare, Nietzsche, a rapper, or the sacred scriptures. Right now, the format and the style of a book, a movie, a series, or a cartoon are defined by the author. It might fit your taste, or it might not. As I have been working on my book, I have realized that I could actually provide multiple versions of it. A geeky technical version, or a more simplified one. A version with lots of dialog, or one with lots of action. A very detailed one, or one that gets straight to the point. In a way, we already do these things. Shakespeare has been rewritten into modern English and even children's versions. In the words of GPT: “Ah, the tales spun by the bard, Shakespeare, Their retelling doth not cease to make air From modern tongue to child’s simpleness, They’ve been spun anew with artfulness. The stories, once told in Elizabethan rhyme, Now find voice in prose that’s most sublime, And even in form most puerile, Their essence doth remain most fertile. So, let the retelling of these tales not end, For in each adaptation, a new story doth bend. And though the words may differ from the first, The themes of love, life, and loss, doth still burst.” While ChatGPT can not yet create pictures, its cousin Dall-E can. And you can combine them by asking ChatGPT to describe characters and use that description to have Dall-E generate them. For example, I asked ChatGPT how would she (Yes, to me is a she) describe an advanced AI persona. This was her answer: Human face with augmented elements: A human face with added technology such as circuitry or digital elements, symbolizing the fusion of man and machine. Next to a Futuristic robot: A sleek, metallic robot with advanced technology and artificial intelligence capabilities. Which I pasted into Dall-E and got the following: It does not take a leap of imagination for a future movie director to use these prompts to create multiple versions of a movie to match the likes of the viewer. So in the future, your Netflix might show the same film to you and your daughter, but hers will be more girly and relationships focused while yours be more techie and facts based. Academic Papers and Academia In General I am currently collaborating on the writing of a paper for a conference. As with presentation slides, a paper also has a similar structure: A long exact title, a set of authors, an abstract, an introduction, a literature review section, the main body of the paper presenting the findings, and a set of conclusions and future work. Papers also need useful graphs and figures. Stats or flow diagrams could be generated using the techniques mentioned above. Interestingly, if you ask ChatGPT to generate a graph, it will give you the code to generate it yourself using Python. An interesting meta prank would be to use ChatGPT to cocreate and submit a paper to a conference on AI models, just to see if it gets accepted. Arxiv would probably accept it. Again, here the benefit of using an AI model is that results could be presented faster. Chat GPT will not yet want to write a paper on a given subject, this is the role of the researcher. Yet, it can help immensely in writing a draft of the findings that the authors can use to tweak and focus on the technical details of the solution they are trying to describe. Conclusion Seeing ChatGPT in action has felt to me like the first time I saw a web browser, or realized I could surf the Internet from the tiny screen of my Palm Treo device. It is a paradigm shift that will change the world as we know it in many ways. We can already see some of its effects like in academies, where they have to AI-safeguard exams by having to move away from “essay-based” grading. What other areas will be affected and how is still an open question. But after having used these technologies for a month in a myriad of situations, I can see how they can improve my productivity in unexpected ways. To conclude this article, I decided to ask GPT how it thinks it will impact the world. Her answer is: The advanced AI-based conversational models like myself have the potential to greatly change the world by improving human-computer interaction and enabling more natural and intuitive communication with technology. In various industries, they can assist with tasks such as customer service, improving accessibility for people with disabilities, and increasing efficiency in data analysis and decision making. These models can also be used to develop personalized and engaging educational and entertainment experiences. However, it’s important to note that with any new technology there are also potential challenges and ethical considerations that must be addressed. For example, there are concerns about job displacement and privacy issues that must be considered as AI-based conversational models become more widespread. Overall, the impact of advanced AI-based conversational models will depend on how they are developed, implemented, and regulated, and it is up to society as a whole to ensure that their use aligns with our values and benefits humanity as a whole. It is interesting to read the convolutions that the model is doing in the 3rd and 4th paragraphs. It feels like my friend GPT is not just an advanced AI conversationalist, it is also its own legal and ethical department, as well as apt at washing its hands for the havoc it can break in humanity. How has your experience with ChatGPT been? In which areas have you used it?
