How to have a better relationship with AI

Five suggestions to shape how you use AI tools.

Introduction

Eric: Hello to readers of The Intelligent Friend. When

Riccardo Vocca

invited me to write an issue with him, I jumped at the chance. The Intelligent Friend has quickly become one of my favorite Substacks and I’m delighted to be here. At my Substack, New Things, I write about our relationship with new stuff in our lives - everything from how to create a habit of mindfulness to the need for calmer technology to the joy of returning to familiar things. I’m also writing a book about how to reshape our relationship with new things to save our time, attention, money, and overall happiness. AI is undoubtedly The New Thing Of The Moment™. So, today, we’ll be offering some tips on how to build a healthier relationship with it. The premise here is similar to a piece I wrote earlier in the year for The New York Times about our relationship with phones (if you’d like to read it, here’s a free, gift-share link 😉).

Riccardo: Having written an issue with

Eric Athas

is a privilege, as is having met him. Thanks, Eric! This article started from reading his writing for The New York Times and from there the collaboration became more and more enthusiastic and stimulating, until arriving at the conception of this piece. First of all, we are, of course, aware that there are many sources and authors - much more authoritative and informed than us - that describe ways to approach AI in a profitable way and so on.

Therefore, to write this article, we started from two pillars. The first is to ‘build’ a piece that is based above all on personal suggestions and lived experiences, and try to integrate it with findings from scientific research. Therefore, not everything we say applies to all experiences, but we hope it can be useful, stimulating, or inspiring for many people who have not yet used these tools.

The second pillar is that this article, although possibly addressed to everyone, is born especially for those people who have not yet used - or have rarely used - Generative Artificial Intelligence tools (such as ChatGPT) or who have interacted in a completely unaware mood (as often happens) directly with AI-based tools (customer service chatbots and so on).

I repeat: we hope that our small contribution can be stimulating and we would be really excited to hear your opinions in the comments!

1. Experiment with AI

Eric: With AI (and any new thing), we often hear about the extremes. On one end, there are people who are adamantly against it. On the other end, sit the people who are all in. I’m here to say that there is a middle ground. You can, and should, find lightweight ways to experiment with AI. Experimentation allows you to dip your toes into AI tools in small, contained tests. Try different tools. If you find something is useful, give it some more time. If you find something is useless, just stop doing it.

I’ve written before about the value of starting small when you’re implementing something new in the workplace. Small-scale pilots at work give you the freedom to test something out without committing to it indefinitely. With AI, I’ve been taking this approach while working on my book. Because I have to do a lot of research for it, I’ve experimented with using AI as a research assistant. I’ll ask a chatbot to find me articles and papers or to suggest a list of experts on a topic. The results are mixed, and always require additional vetting. But these small-scale experiments have allowed me to test the limits and find what works best.

Riccardo: The role of experimentation and the ability to understand which tasks can be best performed with AI is a crucial topic widely covered in the renowned paper “Navigating the Jagged Technological Frontier”, in which Dell'Acqua and colleagues (2023) identify how "the capabilities of AI create a ‘jagged technological frontier’ where some tasks are easily done by AI, while others, though seemingly similar in difficulty level, are outside the current capability of AI". Furthermore, as explained in one of his issues by

Ethan Mollick

, the authors indicate two distinct approaches to collaboration between consultants (subjects of the experiments of the paper) and AI: Centaurs and Cyborgs:

1. Centaurs divide tasks between themselves and the AI, delegating activities according to each party's strengths, effectively leveraging human intuition and AI processing power;

2. Cyborgs fully integrate the AI ​​into their workflow, interacting continuously with the technology, making the AI ​​an inseparable partner with constant interaction and feedback. These approaches offer flexibility in utilizing AI to enhance human abilities and adapt to different work needs and objectives.

In addition to the importance of experimentation, someone could rightly argue that the first fundamental step is to understand, in effect, what to do to start experimenting with AI.

Among the many interesting sources on the topic, an article by the Harvard site suggests the following, intriguing, uses:

1. Creating new content;

2. Generating or debugging code;

3. Creating spreadsheet formulas;

4. Search;

5. Summarizing documents;

6. Synthesizing information, including survey responses;

7. Reviewing or enhancing your work.

2. See it as a tool

Eric: One tip I shared in my article for The New York Times was to make the technology work for you. This came from Google’s productivity expert, Laura Mae Martin: “it’s the intention behind how you’re using it that really makes the big difference.” This applies to AI, too. It’s easy to get carried away by the hype (just look at the Google suggestions below). And I’ve written before about our tendency to frame the present moment in major historic terms. But right now, AI is a tool. You can decide how to use it, not the other way around. Like this Pew Research Center report notes, “technology is neither inherently helpful nor harmful. It is simply a tool.”

Riccardo: I think that among the things I read that struck me most in considering the possibilities that AI as a tool can open up, there is certainly the book 'Co-Intelligence' by Ethan Mollick which, among other things, outlines 4 fundamental principles for working with and using AI. Among these, one I find particularly important for this second suggestion for having a better relationship with AI:

treat AI like a person (but tell it what kind of person it is).

It might seem counterintuitive compared to what Eric just said, but in the vision of this article, it is not. Eric refers to a first approach to AI, and the vision of this as a tool above all to not raise barriers that could jeopardize the improvement of our results - or simple experimentation - with AI. However, when it comes to sitting in front of ChatGPT and writing, perhaps following one of the uses suggested above, Mollick's principles help a lot and make you take a proactive perspective. It is important, as specified by the Wharton Professor, to tell the AI ​​what 'person' it is: is it a journalist? Is it helping you write, so is it an editorial assistant? Is it a chef? And so on. It is a way to proactively orient the approach and interaction with these tools.

3. Try to have fun with it

Eric: When I first started playing around with DALL-E at home, I found my boring prompts were resulting in boring imagery. Then I had an idea. I called in my kids - much more imaginative than I am - and posed the challenge to them: Come up with a scene in your imagination and tell it to me. For the next hour, we had a blast generating vivid images. It was also a chance to educate my kids about AI. Yes, there is a lot of talk about the potential dangers of AI. We should pay attention to those warnings. But it’s also okay to have fun with it as a way to learn about it.

Here was the prompt: A dinosaur, specifically a blue allosaurus with red eyes, crazy purple hair, scary sharp yellow teeth, a long neon tongue and three sharp claws eating pizza with ice cream on it in a candy land with a chocolate ocean in the background. The sand is made from chocolate crumbs.

Riccardo: In my search for papers and possible scientific insights that could give even more relevance to Eric's advice, I found little on the relationship between fun and AI use, except in one specific field: gamification and gaming in general. However, some studies from these fields caught my attention. For this third point, I'll mention one paper in particular that I think is intriguing, published in the journal Psychology & Marketing. The study by Butt et al. (2021) examines how AI tools are transforming the gaming industry by enhancing the consumer experience.

In particular, the study investigates how factors like perceived ease of use, usefulness, advantage, compatibility, enjoyment, customization, and interactivity influence gamers' intentions to use AI avatars. A survey of 500 Chinese online gamers tests these hypotheses. Results strongly support the adoption of AI-powered avatars, emphasizing the importance of customization, interaction, and assistance in providing an enjoyable experience. The study suggests integrating AI into gaming to create a more immersive experience.

Apparently, naturally, this study does not give any real results regarding enjoyment as such, but I think that the concept of an 'AI-powered avatar' that can give a significant contribution to an enjoyable experience is an insight that should not be overlooked and possibly tested in other fields, from everyday uses to those for work.

4. Learn the basics

Eric: Let me be clear: I am not an expert in AI. Not even close. But I am trying to learn about it. With any new thing, knowledge is power. You won’t know whether or not something is good, bad, useful, or useless until you understand the basics. Again, you don’t need to become an expert, but education can come in many different forms. Get to know the basic vocabulary. Read the latest AI news. Follow AI leaders. And subscribe to newsletters like The Intelligent Friend. There’s no shortage of information out there.

Riccardo: Knowing and learning about AI – at multiple levels – could succinctly be defined as 'AI Literacy’. As specified in a fascinating review by Pinski and Benlian (2024):

“Whereas multiple AI literacy definitions exist, they all refer to some dimension of human proficiency (e.g., knowledge, skills, competencies) regarding different subject areas of AI (e.g., AI models, ethical implications, knowledge representations) that enable purposeful usage of AI or interaction with AI".

This is why, simplistically, we could refer to the concept Eric was talking about: starting to learn the basics of how AI works and getting informed about it can be a great way to start interacting with this technology and building a positive and vigilant relationship with it. The relevance of AI Literacy has been expressed, in addition to several academic contributions, also by the World Economic Forum (2022), which, in an article with a rather explanatory title 'Without universal AI literacy, AI will fail us', stated: "Building an AI-powered society that benefits all requires each of us to become literate about AI". We in this suggested fourth point look at AI Literacy primarily from an individual point of view and as a way to begin to reduce the barriers of misunderstanding concerning this technology.

5. Be aware of your comfort level

Eric: For most of us, the stakes are low right now. There’s not a ton of pressure to adopt AI. So, be aware of your comfort level. Play it safe. If you’re uneasy about using AI for a critical work assignment or personal task, don’t do it. You know those stories about AI getting people into trouble? Those are unfortunate but serve as a lesson to exercise caution in these early days.

Riccardo: There is a deep and at the same time intuitive piece of advice that I think fits deeply with this last point of our article, coming from an article by Westerman et al. (2024) on HBR, and it is:

"Start with the problem, not the technology. Wielding a (generative AI) hammer, everything starts to look like a nail. But, instead of asking how to do generative AI in your company, ask what you need to accomplish".

We often think about how we can use AI personally or in a company, to improve productivity. We should, according to the authors, have an inverse approach: we should start from a relevant problem that we often interface with and build a solution by integrating AI into the process. This paradigm shift, as apparently simple as it may seem, actually constitutes a turning point capable of modifying our perspectives and informing the following steps of building and interacting with AI differently.

A final note

Eric: Thanks for reading and I hope you found the advice helpful. And thanks to Riccardo for inviting me on here. As we make our way through the AI hype cycle, here’s some context that helps me: we’ve been here before. Not with AI, but with the rise of social media, the mobile revolution, cloud computing, wearables, search, and much more. We can look back at past new things and learn from them, or rather, learn from how we approached them.

Riccardo: Thank you so much for reading this issue and for your time. I want to sincerely thank Eric once again for collaborating, it was an honor to write a piece with him. I also want to thank all the current readers of The Intelligent Friend and those who read this issue. If this piece intrigues you, I sincerely recommend Eric's newsletter (and all the other things he writes!) about the relationships we build with new things (a fascinating topic).

If the topic of AI through insights that start from scientific papers interests you, well, also stop by The Intelligent Friend, the newsletter in which I try to share everything I learn from reading scientific studies about the psychological, social, and relational aspects of AI.

Thanks again!