Let's talk - a long-form conversation with ChatGPT.

My long-form conversation with ChatGPT...

I had been planning to set some meaningful time aside to put ChatGPT through its paces. Which I did today. To be clear, this was as much about testing my own understanding of the potential of ChatGPT, and the challenges surrounding AI in general, as it was about testing ChatGPT itself.

My plan was to have an in-depth conversation with ChatGPT about an area in which I have significant knowledge already - improving places. In just over 2 hours we explored this in-depth, iteratively. It is hard not to be impressed with ChatGPT's first response, and this is where most people stop and shout ChatGPT's praises, or run for the hills from our future AI overlords. I wanted to dig deeper. Just how far could I go? As it turns out, impressively far.

I'll be honest, ChatGPT's first response was both brilliantly articulated and the typical high-level response I have become accustomed to hearing. Both brilliant and worryingly lacking. Subsequent responses however, in aggregate, accurately summarised the best in current thinking, with a few generic points made, which when I reasoned on their importance, ChatGPT was able to adapt or surface alternative perspectives, data, insights, approaches or solutions - ChatGPT seemed to reason well. Like peeling an onion, each time we stepped deeper, I sought to get to a point of completion which presented a feasible and acceptable understanding of the layer at hand. I found this exercise extremely useful as a means to clearly articulate and clarify my own thinking. We iterated conversationally. Eventually, we hit a point where ChatGPT was reiterating the same summary, in-depth reasoning, explanation of challenges, and bulleted "next step" actions, which we'd got to together. But which still overlooked key challenges, and to my mind solutions to those challenges. Which in itself was useful because in doing so, it identified (potentially, given the limits of what ChatGPT was trained on, and the balance between reasoning created through "our interaction" versus that which was formed from existing ChatGPT "knowledge") considerations and thinking that is new relating to improving place.

Of course, in interacting with such powerful tools we are ourselves acting as a filter and playing a refinement role in that knowledge, and as an extension, the expertise therein. In fact, in time, we will be contributing to that knowledge simply by interacting with it. Simply by reasoning why something did not work, means that the tool becomes more powerful. But I believe that the focus this enables will create value that outweighs challenges.

Key Takeaways:

• Hugely impressive.
• But hugely impressive from answer 1.
• In fact, ChatGPT is so good at articulating responses that most people might not engage further.
• I disagreed with ChatGPT, so felt compelled to carry on.
• Like most things, the more effort you put in, the more value you'll get out.
• ChatGPT is massively useful for distilling "best practice thinking" and "own thinking to date".
• The opportunity with ChatGPT really sits with creating additional layers of "intelligence" to "dig in" on behalf of the user (to put the extra effort in).

?#creativedirector #ai #design #strategy #chatgpt?#longformconversation?