Episode 9: ChatGPT One Year On

Want ideas on what marketing professionals actually think about AI and how they're using it? The AI with DMI series brings you sharp insights from our experts in their fields.  

In this episode Will Francis chats with Clark Boyd about where ChatGPT stands at the end of 2023, a year on from when it splashed on the scene. Has it delivered? Do people know how to use it? Are their jobs in jeopardy? And are companies instructing their teams to use it with clear understanding?  

Check out all episodes of AI with DMI and our digital marketing podcast.

“The more I understand what these tools do, the more I realize how powerful they are not: far more than I did six or nine months ago. And I'm more certain than ever that my job is safe.” Will Francis

Video Transcript

Will

It's been pretty much a year at this point since ChatGPT hit the world over the head and said, AI is here. We've, well, it's been here for a while, but we've finally put a glossy enough interface on it for everyone to care. Are things changing now? Do you think we are at any sort of inflection point? How things matured? Have things gone the way you expected? Where are we now in late 2023 in terms of AI, do you think?

Clark Boyd

It certainly haven't gone the way I expected. They rarely do, to be honest. I make a lot of predictions.

Will

That crystal ball is just... Never very reliable, is it?

Clark Boyd

Pretty cloudy, yeah. No, no, absolutely not. I think the only thing is if you put yourself into the shoes of people who would be making decisions about this technology, especially at that kind of corporate level, then I think it is slightly more predictable because it's fine as soon as something like ChatGPT comes along, someone like me without really that much skin in the game to say, just adopt it. Adopt AI. It's the future.

If you are running a massive company, and that would mean a reskilling of your organization, restructuring, all of these things, you want to make sure this thing is for real. So it has been a period of people running smaller little tests, maybe getting things wrong, but within the confines in which they're safe to do it. And now reaching a point where there's a consensus, right? It's like in the animal kingdom, right?

No animal wants to be the first one to go swimming into the ocean just to see if there are any sharks there, right? There's a hierarchy and people, animals actually do it for lots of different reasons. And then they wave back and sort of saying, it's okay in here if they survive. You know, it's fine. And people want that with things like AI. You don't want to be the first one to take the plunge. There's not actually that much benefit beyond PR, right? You know, you can put something out there and say, X company is the leader in hotel management or whatever. And that's fine. But what if you get AI wrong? You've just wasted a load of money. So.

We've been waiting for lots of big companies to make big investments. So people like Accenture and EY investing billions of dollars into this. IBM investing two billion into global training for people. And that makes people feel like it's safe to go in the water. Right. And this is happening. It's urgent. We've got to go about it now. Intriguingly though, I feel like across the board, it would almost be like a government initiative to bring up people's level of AI literacy before they're even making those decisions. It's been interesting for me because I've done a lot of workshops this year. And I'll be honest, there's been a huge amount of demand, certainly compared to the usual, for me to go out and run workshops for people. And I'm no PhD in AI or anything like that. I've just been working with this stuff for a while and have a decent knowledge of it. And they're like, no, come out, explain this to us.

And as I've lowered the level of sophistication or detail in that, it's gone much, much better throughout this year because I've been meeting people at the level they want to kind of learn about this from. I started it by asking things like, how much do you know about this? Do you know about this or that? But the answer was always no. So I just started with the assumption that, OK, everyone's heard of ChatGPT but do they know how to use it? And people are still talking about things like prompt engineering, right? We just need to get the prompt right, isn't that right? And that's become received wisdom. My challenge with that sort of thing and what suggests to me that a year on we haven't really learned the fundamentals beyond the surface of this, is that, well, we have Dall-E 3 now that is so much easier to work with. It doesn't require these detailed prompts, right? So what are all those prompt engineers going to do? You just scrub that from your CV and move on to the next thing? But if you're just moving from gimmick to gimmick, or like what works today to what works tomorrow, that's very hard, right? It's really hard. You're just running to stand still.

Will

Yeah.

Clark Boyd

Where I'm not necessarily seeing too many organizations yet, and I'm hopeful that this will be the case, is going beyond that to think, well, beneath the surface, what do our people need to know so that no matter what the change is tomorrow, because there will be a change tomorrow, that they're agile enough to be able to work with it so that they understand that this is a new large language model. Therefore, it probably works on the same basis as chat GPT. It's just a faster version, right? Because the surface of it will look different. They'll call it radical, they'll call it revolutionary. They're here to sell.

But the stakes are too high for people to be confused by a lot of this. I'm speaking to people who are both working within companies and the leaders who are booking the workshops, and they'll all openly admit they don't really get what's going on, that it's a little bit scary, but they're under pressure to do something about it. And I've heard managers threatening their teams, saying, well, if you don't sharpen up, 50% of you will be replaced by ChatGPT next year. And the people on the team speak to me and say, we don't think that's true or fair, but we don't know why it's not true or fair because we don't know how ChatGPT actually works and it seems like it can write really good content. So the demystification of this technology is good for everybody except the people selling it, right? Making it some mystical technology from the future is great if you're selling it. That's not a problem. They can do that. But for the everyday worker it's not a case anymore of AI is just something you can leave to the tech team or the nerds will sort that out. I'll go about my business. Now it's in everyone's interest to get a lot more value out of this. And it starts with understanding it.

Will

True that, I mean, it sounds like you've had similar experiences to me. I mean, I personally find that the more, having become like a regular user and really understanding these tools like ChatGPT and Claude and Mid Journey, now that I kind of understand what they do, I realize how powerful they are not, far more than I did six, nine months ago. And I'm more certain than ever that my job is safe.

for now anyway, I mean, who knows what the next iteration is. And I think that's interesting, you're very right, the more people understand about it, the more you understand really what it's good at and what role it plays and how that role slots in somewhere rather than just replacing everything. So that is very interesting. But as you say, we're always working with the worst version of it. This is just a preface to the next thing that comes along.

Clark Boyd

Yeah.

Will

I mean, is the next thing that comes along going to be any different in that regard?

Clark Boyd

If you listen to the people that are building a lot of this stuff, then they would have you believe that, yes, it will be radically different. I'm always a bit skeptical on those things. I'm always looking at what they've got to sell, what share price they're trying to drive up by saying these things, who they're trying to impress, what their promotion is based on, all of that stuff. Because we did, of course, have people saying, this time last year...

There was a lot of talk about, well, ChatGPT is here. It's not good at this, and this. But let's not talk about those things because it will be good at them probably. Everyone's saying it will be good at those things. And some of those problems have been here to stay, right? There's a kind of saturation point. There's a point that you really struggle to get beyond. It can't be exponential growth forever. We'd love to think it is. And what we hear from people that are working on this stuff very directly is that the future is going to be more about AI systems talking to each other. So people putting them to work.

And we've been talking earlier about how Google is integrating this within all of its apps, right, so that you could get things done as long as it's within the Google ecosystem by simply asking the assistant and it will go and fetch things from different places, move those the research that it does on your market into Google Sheets and then run a visualization of it, put some of those visualizations into Google Slides for you and then even email it off to somebody if you're happy with what it's done.

that all of that sort of thing is just the AI talking to different versions of itself. And with, and this exists already, Agent GPT will let you do this. There's one version that is the leader and it creates sub teams and gets them to go and do things and it puts in checks and balances as well, breaks everything down into micro tasks and goes and gets it all done for you. There are challenges with that, plenty of challenges in terms of the quality of the output and how you really integrate that with the way people work already.

I think given that that's the direction of travel, how fast we get there, we don't know, but it's definitely the way things are headed. It still comes back to this point about people understanding what it is that the technology can do and how it can help them get done, whatever it is they want to do, whether it's faster, cheaper, better, they need to know that. And the person that was saying this, one of the co-founders of DeepMind, he was saying, you know, the future belongs to people with ideas, people who know what they actually want to get done, people who are intentional about things. And I think so much of working life, I certainly know from my own experiences is going through the motions, right? I need to get that spreadsheet done, I need to get this to that person and the days go by and people are busy, right? We're meant to be less busy because we've got all this technology. Productivity is stalled, productivity is completely stalled, especially in the UK, it just hasn't gone forward. And yet we're led to believe that we're living in the future and everything's, you know, jetboats and it's all AI.

Related Articles

• AI in Digital Marketing - The Ultimate Guide
• ChatGPT Prompt Guide for your Digital Marketing Plan