Let's kick off this series with undoubtedly one of the most intelligent people using ChatGPT.

TBH, I had a little doubt but only insofar as whether to choose Stephen Wolfram whom will follow so you should subscribe to not miss it – or David Deutsch.

So the races were of between two smart Brits whose accents are as enjoyable as their intelligence they use to make this world a better place.

Practicality won and I just recently have listened to the “Deutsch Files” where David Deutsch talks about his adventures with ChatGPT and generative AI amongst many other fascinating things. So, head over if your curious, but before that let’s go in medias res.

David Deutsch’s View on ChatGPT

“it's a phenomenal chatbot. I thought it would be decades before we had a chatbot that good,” a view David Detusch shares with Stephen Wolfram.

But “it was never intelligent,” he insists. This is not AGI nor ASI nor will it ever be.

How David Deutsch uses ChatGPT

I'm just using ordinary ChatGPT

—David Deutsch

David is using ordinary ChatGPT but without any plugins saying, “now there are a bunch of plugins, but they haven't really worked for me so I'm just using ordinary ChatGPT.”

David Deutsch’s Prompt Engineering Tactics

“and if you're very good at making the prompts - which I'm not yet so maybe I'm underestimating it. But, the better you are at making the prompts the more it will tell you what you wanted to know.“

—David Deutsch

“[The] better you are at making the prompts the more it will tell you what you wanted to know,” can be interpreted in two ways.

David's expereince underscores the critical role of prompt engineeering in harnessing the full potential of ChatGPT.

The art of prompt crafting is not necessarily one he is master of yet as he candidly admits, “and if you're very good at making the prompts - which I'm not yet so maybe I'm underestimating it. But the better you are at making the prompts the more it will tell you what you wanted to know.”

The effectiveness of ChatGPT, and indeed any AI, is not just a matter of the technology's design but also how we, as users, engage with it. It underscores the idea that mastering prompt crafting is not merely a technical skill but an intellectual one that significantly influences the output you receive.

While this might soon change, for now there is still the necessity for us users to develop a certain level of proficiency in interacting with AI to fully leverage its capabilities. This is a reminder that while AI tools have become more accessible, their most effective use requires a learning curve and an appreciation for the nuances of human-AI communication.

AI as a collaborative tool

Additionally David speaks to the potential of AI as a collaborative tool. When Deutsch mentions, “the better you are at making the prompts the more it will tell you what you wanted to know,” he is pointing out that AI's utility is not static. It evolves with our ability to communicate our needs and questions effectively. This dynamic suggests that AI, in this case, ChatGPT, can become an increasingly powerful tool in intellectual explorations and problem-solving, provided we learn how to interact with it properly.

If you find yourself exploring ChatGPT’s possibilities, don’t lose confidence. Even experts in fields as complex as physics find themselves on a learning curve when it comes to maximizing their use of AI tools.

The key to unlocking the full potential of AI lies in our hands – or more precisely, in our prompts.

And it will take you several attempts to just get it right, especially if you want it to achieve a little more than just the first plausible sounding answer. Which brings us to another prompt engineering tactic: iterative prompting.

Iterative Prompting for Complex Questions

For a complex question it usually takes me two or three goes to correct it.

—David Deutsch

The quote from the earlier chapter continues as follows, "for a complex question it usually takes me two or three goes to correct it. And sometimes it just won't correct it."

Because then there is the iterative nature of working with generative AI to achieve desired outcomes on top of getting the prompt right.

These experiences and challenges with that and the intricacies involved in prompt engineering David Deutsch details in his anecdote using DALI to generate an image of Socrates, Plato, and their contemporaries.

The current capabilities and limitations of AI technology, especially in understanding and executing complex, nuanced prompts can be quite frustrating.

Let’s observe a real-time exploration of the iterative nature of human-AI collaboration by following David Deutsch as he navigates through the process of refining his requests.

And sometimes it just won't correct it.

For example just yesterday I asked it to produce a picture with the DALI plugin. There's a picture that I had wanted for my book, but which couldn't really get an artist to draw. But if I had my previous book again, I would want a picture of Socrates and the young Plato and Socrates's other friends all sitting around. I said make me a photorealistic picture of that.

So DALI made a black and white picture and I thought, okay I can't say that it's not photorealistic but I meant color photorealistic.

Also, it had Socrates sitting in a sort of throne and everybody gathered around him. So I said put Socrates down the same level as everybody else and by the way make Plato a bit taller, even though he's a teenager but he's a wrestler.

So the next output Dali generated pictured Socrates down, but still taller than everyone else even though I told it not to do that. And Plato was sort of topless and sort of ripped and with muscles. So now he was a wrestler when I just said he has a wrestler's build, which is what I call him in the book “Beginning of Infinity” because nobody knows what Plato means and it was a nickname. But it may have been that Plato means “platon” which means broad. And he was a wrestler, so you know put two and two together. And he had a broad build like a wrestler, too.

But then from then on, I tried three or four more prompts and I just couldn't get it to clothe Plato again after it had got that wrong the first time.

I couldn't get it even though I explicitly told it.

So the functionality is tremendously good. And the first black and white picture DALI produced was pretty impressive. And if I hadn't told it, or I should have thought to tell it not to make Socrates stand out among the others…. But then it got down the wrong track. And I don't know how to make it not do that.

You can personalize your prompts. I tried doing that.

And it made it worse than before.

Deutsch's experience highlights a critical aspect of engaging with AI: the need for precision and clarity in communication, and yet, the simultaneous challenge of the AI's literal interpretation of instructions. His initial request for a "photorealistic picture" resulted in a black and white image, which, while technically meeting the criteria, missed the implied expectation of color. This discrepancy between Deutsch's vision and the AI's output underscores the importance of explicitness in prompt crafting.

Moreover, Deutsch's subsequent attempts to adjust the depiction of Socrates and Plato reveal another dimension of AI interaction – its difficulty with nuanced adjustments. Despite several iterations, each with more specific instructions, the AI struggled to align with Deutsch's envisioned outcome, particularly in regard to character positioning and attire. This persistence, alongside the AI's unpredictable interpretation of prompts, reflects the complex interaction between human intent, a text chat GUI and AI execution.

Deutsch's anecdote is a situation I find myself in often. One the one hand there is this impressive capacity to generate outputs from my vague prompts, and on the other hand it ChatGPT is faltering when faced with the subtleties of my human expectation.

Precision is bounded by the clarity of our communication.

It provides a tangible example of the iterative process necessary to hone AI's utility, showcasing the blend of patience, specificity, and creative troubleshooting that defines effective human-AI collaboration.

Deutsch’s story serves as a reminder of the ongoing journey we all experience toward the seamless integration of AI in our creative and intellectual endeavors.

This mix of fascination and frustration is basically our entire interactions with AI tools and every tool that we want to use to get tasks done.

I can just keep encouraging you to keep a mindset of exploration, adaptability, and continuous learning.