Unwrap Savings: Save 20% on all courses this December. Ends 14/12.

Unwrap Savings: Save 20% on all courses this December. Ends 14/12.

Find Your Course

Case Study

What does ChatGPT mean for the future of work?

ChatGPT has set the record for the fastest-growing user base in history. Within just two months of its public launch, it reached 100 million active users. No other application has come close to that level of huge, instant popularity. It reached the 1 million user milestone after 5 days, compared with the leisurely 75 days it took Instagram to reach the same mark.

ChatGPT’s creator, OpenAI, was more surprised than anyone by this level of success. After all, ChatGPT is simply a user interface that allows us to query OpenAI’s core technology. It is what is known as an ‘API wrapper’, because it bridges the gap between an Application Programming Interface and an end user. In other words, it allows everyday, non-technical users to access sophisticated technology.

Other companies, including Google and Meta, have also invested heavily in this area of artificial intelligence (AI). They use large language models (LLMs), a type of AI that powers new technologies by processing and understanding vast amounts of natural language data. They are trained using machine learning techniques on massive datasets of text, such as books, articles, and websites.

During training, LLMs learn to identify patterns and relationships within the data, allowing them to generate new, original text that mimics human language. This means that LLMs can be used to generate a wide range of content, from simple sentences to full-length articles, stories, and even dialogue. The answers we see are similar - but not identical - to language that has been used before.

So why did generative AI take off for OpenAI first?

The seeming simplicity of ChatGPT is its killer application. As with the Google search query box, the user must enter their request in detail before the technology kicks in. This stimulates the conversation using a rich data source.

We now have at our fingertips a technology that can instantly create plausible content on any topic. The quality is variable but it will only improve over time. The quantity is endless, in theory. 

Businesses and professionals are rightly curious about the implications of ChatGPT, and similar generative AI tools, for the future of work. The common questions I hear include: Does ChatGPT mean that AI will replace people? How rapidly will it advance into other areas of work? What skills does the workforce of the future need?

We are still processing the strengths and weaknesses of the technology, but early experiments illuminate some possible ways to understand the future. A new study from MIT (“Experimental Evidence on the Productivity Effects of Generative Artificial Intelligence”) finds that when writers use ChatGPT, they can produce both a higher quantity and a higher quality of work. 

In the study’s experiment, writers carried out “mid-level professional writing tasks” including emails and market reports. These are structured tasks that require little context or creativity, but the results still show that a positive collaboration between professionals and ChatGPT is possible. 

This outcome is far from inevitable, however. The trick is for writers to know how and when to use ChatGPT for assistance. 

The experiment found that if writers asked the technology to write full articles, the quality (as graded by experts) was relatively low. This will ring true for anyone that has asked ChatGPT to write a blog post. Nonetheless, they were able to produce a huge quantity of this mid-quality work. 

When the writers used the technology as a partner for brainstorming, structuring, and drafting the piece, they could focus their time on finessing the ideas and then editing the content. This was still quicker than writing the articles from scratch and the quality was graded much higher than content created by ChatGPT alone. 

As the authors conclude:

“ChatGPT substantially changes the structure of writing tasks. Prior to the treatment, participants spend about 25% of their time brainstorming, 50% writing a rough draft, and 25% editing. Post-treatment, the share of time spent writing a rough draft falls by more than half and the share of time spent editing more than doubles.”

Will this pattern be replicated in other areas of work?

Writing is just one of hundreds of tasks we complete at work, yet the same division of labour will likely apply elsewhere. If we rethink our work from scratch and separate out the specialisms that are best carried out by either people or technology, we can deliver similar productivity gains. 

This realisation led The Atlantic to describe talking to AI as “perhaps the most important skill of the 21st century.” Those who understand the fundamentals of the technology will use it to outperform those who work alone. New job roles have opened up already, most notably the “prompt engineer” responsible for crafting the text prompts that encourage ChatGPT to deliver the desired replies. 

For my part, I was able to ask ChatGPT to summarise my notion that everyday tasks can be split between AI and people, to great effect:

We opened this article by defining ChatGPT as an API wrapper that makes OpenAI’s capabilities accessible to a mass audience. Understanding this nuance is key to unlocking the future of generative AI. 

OpenAI took a powerful piece of technology (their own Large Language Models) and found a way to make it appealing to everyday users. First, it had to pinpoint a real user need. Then, it developed a solution that the user could access easily. 

The same possibilities lie ahead for every business and every professional. The technology is open for all to use, if only we have the imagination to put it to a valuable end. 

On March 28th 2023 we hosted a webinar where we explored these themes in more detail, with examples of how to use generative AI for marketing and tips to prepare for the future of work. Follow the link below to see the recording.