In this issue

• OpenAI launches ChatGPT Enterprise

• Storm through your desk research in seconds

• Studies find AI gives humans a 40x creativity boost

• Why we’ve created The AI Writer

AI DEVELOPMENTS

OpenAI launches ‘secure, private’ ChatGPT for organisations

ChatGPT Enterprise includes industry-standard security and privacy features, and comes with a system to allow companies to track usage.

OpenAI appears to have removed one of the biggest barriers to the widespread adoption of AI in the workplace: security.

Early this week, it launched ChatGPT Enterprise, a corporate-friendly version of its popular AI assistant. But unlike the regular version, this one includes enhanced security features, including encryption and compliance with a widely accepted data-management standard called SOC 2.

Privacy worries

As we’ve previously reported, many large organisations have been worried about what their employees are uploading to ChatGPT ever since its launch last November.

According to OpenAI’s own figures just released, workers from 80% of Fortune 500 companies have been using it in their jobs. But, as we revealed last month, tens of thousands of employees have already shared confidential information with the chatbot.

This is a problem, as OpenAI trains its models partly on what people type into it. So it’s possible that ChatGPT could end up sharing proprietary information when it answers queries from other users.

That’s why Apple, Amazon and Samsung have all restricted its use, fearing their company secrets could end up in the hands of competitors.

Worth the cost?

ChatGPT Enterprise is OpenAI’s answer to these concerns, offering not just a secure environment but also information on how it’s being used. The company also says the new, corporate version is up to twice as fast as its most recent, subscriber-only model, GPT-4.

And it can handle twice as much data as the regular, free version of ChatGPT.

ChatGPT Enterprise has already been piloted by several high profile companies, including PwC and Estée Lauder.

OpenAI would not reveal details of what the new service costs when we asked them, saying only that it depends on each company’s use case. Enterprise versions of software are typically much more expensive than those aimed at consumers.

But, given the potential of AI to transform how we work and cut costs, many organisations will likely decide that even a high price is still one well worth paying.

RESEARCH:

AI makes humans 40x more creative

Humans working with AI could generate 800 ideas an hour, uncovering the best ones in record time. Fewer than one in ten humans could match ChatGPT’s level of innovation.

Most advisors, including me, often dismiss ChatGPT as a mere machine. We tend to say that it takes a human to be truly creative. But new research is challenging that view.

40x more creative

One study, for example, found that a human collaborating with GPT-4 could generate approximately 800 ideas an hour. That’s 40 times the rate of a typical human working on their own.

It also found that working with AI was a lot more cost effective. Each new idea cost a mere 63 cents, compared with $25 for those that humans working unaided were able to come up with.

Another series of experiments employed the Alternative Uses Test to measure creativity. This involves presenting objects (say, a ball, or a tyre) to volunteers and asking them to suggest new uses for them. ChatGPT beat more than nine in ten humans (90.6%) in such tests.

And research at the University of Montana found that an AI scored in the top 1% in the Torrance Tests of Creative Thinking, outperforming the majority of college students who participated.

Numbers game

That’s not to say that all AI ideas are good ones. Far from it. But that’s not the point. Creativity is partly a numbers game.

Previous research into how innovation works has shown that good ideas usually exist amid lots of bad ones. So quantity matters as much as quality.

A human using AI can think up a lot more ideas in a shorter time. And while 800 AI-generated ideas will include plenty of duds, a human-plus-AI ‘thinking machine’ will be able to think its way to a clutch of good ideas at a blistering pace.

Together, these studies show that ChatGPT and similar bots have the potential to be much more than just a conversational agent. It could even be that it’s largely our human imagination that’s limiting what we use them for.

ME, MYSELF & AI

Using ChatGPT to analyse data

Adapting this prompt could supercharge your research and save you hours – or even days.

The full potential of AI to make us more effective and successful in our professional roles is often far from obvious.

OK, so it can generate text on almost anything in seconds. It will even rewrite it in whatever style you want (Shakespeare or Che Guevara – you choose).

But these trivial uses hide the fact that it can make working life a LOT easier.

Sector research

Take market or sector research, for example.

In my case, I’m working on a book (on the brain science of reading and how it rules our lives).

It’s what publishers call a 'big idea' book. That’s the sort of thing that authors like Malcolm Gladwell (Blink), Yuval Noah Harari (Sapiens) or James Clear (Atomic Habits) might write, for example.

But to be successful, it needs to focus on the needs and interests of its potential readers. (This is true of all written communication.) I therefore need to find out who reads big idea books, as they’re my target audience.

Perhaps surprisingly, ChatGPT can help with this. Here’s how.

Update the data

First, I need to address a problem that most people have when using ChatGPT for research: much of its knowledge is out of date. It stopped collecting data in September 2021, so it won’t know about anything that’s happened since.

But there’s an easy fix for this, and that’s to feed it data yourself. Fortunately, in the digital age, there’s usually plenty of that publicly available.

In my case, I knew that all the above authors were prolific users of social media. So a sample of their recent posts might give me an insight into the kind of people they might be targeting.

Anyone can do it

To test this theory, I first used this tool to collect the text of the last 100 tweets from one of the authors I wanted to analyse.

This gave me a CSV file that I could then simply copy and paste into ChatGPT after first giving it this prompt:

Act as my market research assistant, Bella. Your job is to help me study the success of certain authors. 

I'm currently studying the author Yuval Noah Harari. In a moment, I will give you 100 of his tweets. 

Your job is to analyse them to deduce his target audience. 

Each tweet will include the number of retweets, quotes and likes. Use these data to infer the popularity of each tweet. 

Then analyse the text of each tweet to determine whom the author was targeting when he wrote it. 

The dataset will also include the number of replies and any text that accompanied quotes. 

It lists who retweeted, quoted or replied to his tweets. Use this information to work out who his posts are resonating with the most. 

Finally, combine your insights to give me your assessment of five personas that the author is targeting, with evidence. 

Do you understand? Feel free to ask me clarifying questions if necessary.

‘Bella’ then replied to confirm that she understood and asked me to paste in the data, which I did.

Then she got to work. Here’s what she came up with, in less than ten seconds:

Tip: ChatGPT restricts how much information you can give it. But this tool will help you break down larger data sets into chunks it can easily cope with, if you want to use a bigger sample size.

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.