Has the AI bubble finally burst?

Plus: the surprising secret of AI success and why I let an AI bot run my life.

Hello, and welcome to Issue 13 of The AI Writer.

The initial buzz about AI is now silent. ChatGPT no longer dominates headlines and dinner party conversations. So does this mean the bubble has burst?

We take a look beneath the surface to find the true picture.

We also reveal the surprising secret to AI success. (Hint: you may even have an advantage if you’re old school and not one of life’s natural geeks.)

And I reveal why I’m now letting an AI bot run my entire working life.

This issue will take you 10 minutes 38 seconds to read in full. Too long? Read the summaries in italics in just 25 seconds instead.

Let’s dive in.

– Rob Ashton

In this issue

  • Has the AI bubble burst?

  • Success with AI depends on being human

  • The AI that now runs my life

  • Why we’ve created The AI Writer


Has the AI bubble finally burst?

Woman in 30s at funfair, dressed in dark blue puffa and bobble hat, smiling and balancing a bubble on her index finger

(Enrico Carcasci / Unsplash)

The initial hype may have died down. But look beneath the surface and you’ll see every sign that the foundations of our working lives are shifting.

It’s sometimes difficult to get to the truth about what’s really happening with AI. For every headline on how it’s going to take over the world there’s another one claiming that users are deserting the technology in droves.

It’s true that the febrile hype that dominated the news cycle early this year has all but disappeared. And the number of new people signing up for ChatGPT seems to have peaked.

Yet focusing on just one statistic can be misleading. A survey of more than 4,000 people found an even split between the number of users and non-users of generative AI apps like ChatGPT.

But the devil, as always, is in the detail.

For a start, usage varies widely depending on where you live.

Fewer than one in three Brits (29%) are using it, for example. But almost three quarters (73%) of Indians are. This compares with 45% of Americans and 49% of those surveyed in Australia.

And those who are using it are using it a LOT. The early adopters are not just users but enthusiasts.

Around half (52%) of people who are using it say they’re doing so more than ever. Three quarters (75%) say they are looking to automate tasks at work and use it for work communications. And many of them believe they’re well on the way to mastering it.

Perhaps not surprisingly, it’s younger workers who are the main early adopters. Two thirds of users are Millennials or Generation Z. Two thirds of non users (68%) are Generation X or Baby Boomers.

Hype cycle

What we’re really looking at here is the classic Gartner hype cycle of tech adoption:

Red line graph, showing visibility versus time. The curve starts at zero (the 'technology trigger'), rises steeply to a peak ('the peak of inflated expectations'), then drops just as steeply to a much lower point (the 'trough of disillusionment'). It then rises through the 'slope of enlightenment' before levelling off at the 'plateau of productivity' below the peak level.

(Jeremykemp / Wikipedia)

First, a new technology bursts onto the scene and everyone is full of excitement and talk of what it might do. Then the initial hype dies down and doubt sets in.

But if the tech then starts to meet or even exceed initial expectations, more people start using it again. Sooner or later, it becomes an accepted, productive part of our lives.

A clearer view

The first half of the year was largely dominated by two types of news story. Either AI was going to upend our entire lives or it was going to destroy them.

Then, of course, there was the inevitable backlash – a rash of reports of how the AI boom was over.

But the clamour for eyeballs on social media can cloud the view of what’s really happening, which is more nuanced.

Peer just beneath the surface and you’ll see companies are adopting niche software built on powerful AI technology. And the tech itself is still developing at a blistering pace. All that’s changed is how much notice the average person is taking of it.

In fact, AI is already transforming some of our biggest – and most conservative – industries. Take the legal sector, for example.

Unlikely champions

Lawyers are not the world’s most natural early adopters. (After all, they’re professional sceptics.) But many large firms are already using AI-based apps to transform how they work.

For example, Allen & Overy, one of the world’s biggest, has been using an app called Harvey to enable its more than 3,500 lawyers to automate tasks such as legal research.

AI is also breaking down barriers that could allow potential competitors outside the profession – such as the Big 4 accounting firms – to start offering legal services.

Those four firms could be very well placed to muscle in, as they’re all placing huge bets on AI.

As early as last December, Deloitte announced that it was budgeting $1.4 billion for AI and tech training.

In April, PwC said it would invest $1 billion in AI over the next three years. Then KPMG trumpeted a $2 billion AI partnership with Microsoft in August.

And just yesterday, EY launched an entire arm of its business dedicated to AI and announced it would train its 400,000 employees in the technology.

New stage

We’re now entering a new stage, where we think less about the novelty of the technology itself and more about what we can do with it. That can only be a good thing.

Look beyond the current news headlines and it becomes obvious that the tech foundations on which our modern lives are built are shifting. Maybe we won’t get an AI earthquake. Perhaps change will happen quickly but more subtly. But it’s still set to transform our lives.

Apple is now reported to be spending millions of dollars a day on making personal, generative AI small enough to fit in its iPhones. Its latest iOS now includes an option to clone your own voice.

The next iPod moment, when we all carry an AI ‘second brain’ in our pockets, is getting closer.

The hype may have died down, but don't be fooled. It’s only the attention bubble that's burst. AI itself is progressing faster than it’s ever done.


The secret of AI success? Be more human

Man in early twenties sat in a classroom, working at a computer

(Desola Lanre-Ologun / Unsplash)

Surprisingly, getting the most from the newest of new technologies depends on three of our most traditional skills: empathy, clear thinking and good written communication.

We’ve been saying for a long time that organisations will make the most progress not by replacing their workers with AI but by using it to bring out the best in them.

Staff and the knowledge capital locked up in their brains take years to hire and build up. So rushing to replace them with a bot risks throwing that away, especially if an AI replacement programme fails to live up to expectations.

Long-term risks

But there are other, more long-term risks too.

Outsource to AI too soon and you could end up automating procedures before the humans have had a chance to use AI to optimise them. And, as bots usually work at a blistering pace, this could scale up bad practice.

A better alternative is to use generative AI apps like ChatGPT for the things we’re not so good at or that take us too much time.

This can get us to a point where we have our best ideas – the real, breakthrough moments – in hours or minutes rather than months it usually takes. (If we ever have them at all, as we’re too exhausted from all the grunt work.)

Best work

Even on a day-to-day level, AI can still help humans do their best work.

As we reported back in July, for example, research has found that bots working alongside customer service agents can increase productivity by an average of 14%. And in the least experienced workers, the improvement was more than double that (30%).

A conversational AI app was able to suggest solutions to agents in real time while their human counterparts were deep in conversation with frustrated customers. The humans could then choose to accept or reject the solutions, many of which may not have occurred to them.

New hires were also able to reach the productivity levels of their more experienced colleagues up to five times as quickly when working with an AI chatbot in this way (in just two months, compared with eight to ten months when working without it).

Thinking for the bot

But there’s one skill that AI bots need from us in order to do their best work: clear thinking.

It takes a human to deconstruct a problem in a way that a generative AI chatbot like ChatGPT can really work with. Prompt engineering (writing AI instructions for the best results), is not just a technical, ‘hard’ skill but a soft one too.

To get the best from a bot, you need to think your way into how its ‘mind’ works and then communicate with it in a way that it will understand.

And if that sounds anthropomorphic, it’s meant too. Because writing for a bot is very similar to writing for a human.

Human talents

It rests on stepping into its shoes and seeing things from its perspective, just as we need to do when writing to our colleagues or customers. ChatGPT can’t do that, but humans can.

We develop the ability in early childhood and it becomes central to how we interact with everyone else. (I wrote about this last year in an article called Mind reading for beginners.)

Success with any AI chatbot depends largely on prompting it – that is, writing to it – in a way that draws on that ability.

In other words, getting the best from the bots relies on those most human of talents: empathy, clear thinking and effective written communication.

Tech developers are forever in search of the ‘killer app’ – a piece of software that becomes so key to success that everyone must have it. It’s that quest that’s driving the AI gold rush.

But the real killer app is not AI. It’s human plus AI.


Why I’m now letting AI run my life

An app called Rewind is now recording everything I do and enabling me to search it with an AI chatbot. What could possibly go wrong?

I’ve got a confession to make.

I’ve gone all in. I’ve taken the red pill.

I’ve jumped into the AI pool with both feet and let it run my life to an extent that I never thought I would.

What I’ve done is not for the faint hearted. It’s going to sound like the stuff of nightmares for any IT manager. To be honest, I was reluctant to do it myself at first.

But now, I’ve installed an app that records everything I do on my computer and powers an AI assistant.

It logs every screen, hears ever word, detects every click. And then it makes all of this data searchable through a personal chatbot.

The app is called Rewind, and when I first heard about it, it really did feel like a step (much) too far.

Frenetic search

I could certainly see the appeal, especially for someone like me.

I have ADHD, and my mind moves quickly. Sometimes, the pace of my research borders on frenetic. And when I’m hyper focused on a particular topic, I’ll follow it all the way down the rabbit hole and back up again.

I won’t always register what I’ve seen, but those snapshots of information frequently resurface and connect when I’m least expecting them. That’s how I've had some of my best ideas.

The trouble is that ideas are not enough, especially when you want to share them in a newsletter that’s read by thousands of people.

I never publish anything without reviewing the original source. And I always check that the fleeting impression I got in my semi-feverish search state was accurate.

But the condition that gives me my fast mind and my hyper focus is also the one that can cruelly rob me of any memory of what triggered them.

Missing threads

I’ve often lost literally days searching for the web page, email or even personal note that sparked and supports a particular insight. More than once, I've failed to find it and been haunted for months by the thought that a vital piece of a puzzle will forever remain stubbornly out of reach.

Obviously I’m not the only one with this challenge. We’re all busier than ever and, even though we all live in front of our screens, our brains haven’t evolved to cope with such a limitless supply of information.

Rewind claimed to be able to find all of those missing threads and weave them together for me again, through a conversational chatbot that could even provide links to the places in which I’d originally seen them – on my computer or my phone.

Meeting summaries

I have to say, I’ve been very impressed so far.

It even automatically produces summaries of Zoom meetings for me. (Zoom itself can now do that too, but this does it automatically as part of my overall workflow.) And it can draw on summaries from several meetings.

I can ask it to sum up, say, discussions about our marketing plan for the last quarter and it will do it.

It’s not perfect yet. Summaries can lack detail (a problem that the company tell me they’re aware of and are working on). So I often have to look at the individual meeting transcripts that it also produces automatically. And those transcripts are by no means the best I’ve seen AI generate.

I never need to worry about forgetting a vital piece of research again.

But for me, it’s still a huge leap forward.

If I ask, say, ‘Where did I read about the use of AI in law firms?’, it will reply in everyday language, ‘You were looking at that on August 28th on this website, and you made some notes on the same idea three days later.

And there will be snapshot images of those sources along with links so that I can go straight to them.

Stored locally

For someone like me, this has been like finding the Holy Grail. But what about the obvious issue: security?

In an era where data breaches are the norm and our digital footprints are larger than ever, the idea of an app recording everything sounded like a privacy nightmare.

That was why, when I initially learned about it, I still chose to rely on my chaotic brain.

I was only finally persuaded to try it when I discovered that all the data that Rewind captures is compressed, transcribed, encrypted, and stored locally. That means most of my information stays with me, not on some distant server.

Powered by GPT-4

Most information? Well, the app runs on GPT-4 – the latest, most powerful large language model behind the subscriber-only version of ChatGPT. So it still needs to send my queries to the cloud. (Sam Altman, ChatGPT’s founder, is also an investor in Rewind.ai.)

As with all third-party apps that use GPT-4, though, my query won't be used to train the AI and will be deleted after 30 days.

When you ask a question, the app first searches for the most relevant moments from your activity and it’s only those that it sends to GPT-4 to generate an answer. (In other words, it’s not uploading your entire personal history.)

It also only sends a text record, not the images themselves. Everything else stays with you, on your hard drive.

You can also specify what it does and doesn’t record, or set time limits for what information it uses. The app also doesn’t record sensitive information, such as credit card numbers or passwords.

Knowledge workers

I’ve been testing it for a couple of months now and I’m already wondering how I managed without it. Not only does it save me hours every week but it also ensures that I never need to worry about forgetting a vital piece of evidence again.

This, to me, is an example of the real potential for AI to transform the lives of knowledge workers. It represents a significant leap in how we can harness AI to improve our productivity and streamline our digital interactions.

It’s not just about recording; it’s about finally bringing order to our digital lives and enabling us to fulfil our potential.

Large organisations will understandably have privacy concerns. So it will be interesting to see how the app develops now that OpenAI has announced a corporate version of ChatGPT.

And most workers are unlikely to get access to Rewind anyway as it’s currently only available for Mac and iPhone. But a Rewind spokesperson told me the company plans to announce news of PC and Android versions soon.

We received no compensation from Rewind for this review

Why The AI Writer?

When I founded Emphasis, 25 years ago, it was just me, a cat and a kettle. It’s since grown to become the most trusted provider of business writing training in the world and helped more than 80,000 people, from 32 countries.

We’ve worked with tech giants, top 10 law firms and major financial institutions, as well as at the highest levels of government. (We’ve even sent trainers to work with clients in the Himalayas.)

I would not claim that we’re experts in everything, as we’re certainly not. But, after working with the authors of around 100,000 documents, it’s safe to say that business writing is something we do know a thing or two about. And that includes witnessing all the various ways in which organisations get it wrong.

Now, we’re at the forefront of using AI to help them get it right.

Of course, I’ve seen nothing like the AI revolution that’s hitting us. Nobody has. But I do feel a responsibility to use our experience to help people navigate this brave new world of communication bots.

So here’s the deal. You get on with your job, while we immerse ourselves in the latest developments in workplace AI. We’ll worry about keeping up so you don’t have to. Then we’ll tell you exactly what you need to know to stay ahead of the curve.

Please forward this email

Feel free to forward this email to anyone who might find it useful.

Each issue of The AI Writer takes days to research and write, so the more people it helps, the better.

And if someone forwarded this to you, you can grab your own, free subscription here.

That’s it for this issue

You can catch up on all previous issues here.

Why not hit reply and let me know what you think of this one? Your message will go straight to my inbox and I’d love to hear your views.

– Rob

PS. I also regularly post new content on LinkedIn. Feel free to connect there, too.