Co-Intelligence: Living and Working with AI

Co-Intelligence: Living and Working with AI
Ethan Mollick
WH Allen, 2024, 234 pages

 

‘Humans are far from obsolete, at least for now’

The speed of progress and the volume of commentary have made it challenging for the busy professional to keep up with developments in generative AI since ChatGPT was released in November 2022.

This also means there hasn’t been time for the more considered scholarly reflections to appear in peer-reviewed journals. Yet one academic has emerged as the go-to voice of reason in this space and he’s now shared his thoughts in this highly readable new book.

Ethan Mollick, an associate professor at the Wharton School in the US, has over 100,000 followers on LinkedIn. He’s our guiding light to help the rest of us understand the implications of AI on our lives and our areas of professional practice.

He’s keen to let us know he’s no computer scientist: rather than being a geek delighting in technology, he’s an academic studying innovation who experiments with ChatGPT in the classroom and who is keen to look to the future. Yet ‘no one really knows where this is all heading, including me.’

Here’s how he puts AI in perspective. ‘We have invented technologies, from axes to helicopters, that boost our physical capabilities; and others, like spreadsheets, that automate complex tasks; but we have never built a generally applicable technology that can boost our intelligence.’ That’s what lies behind the concept of co-intelligence.

The author provides a brief introduction to the long history of AI, written with commendable clarity. He describes large language models as prediction machines (‘ultimately, that is all ChatGPT does technically – act as a very elaborate autocomplete like you have on your phone’.)

We’ve reached the point at which AI can be described as an emergent technology, exhibiting human-like intelligence and creativity. ‘In practical terms, we have an AI whose capabilities are unclear, both to our own intuitions and to the creators of the systems. One that sometimes exceeds our expectations and at other times disappoints us with fabrications. One that is capable of learning, but often misremembers vital information. In short, we have an AI that acts very much like a person, but in ways that aren’t quite human. Something that can seem sentient but isn’t (as far as we can tell). We have invented a kind of alien mind.’

Which brings us to the ‘alignment problem’ – ensuring AI serves, not hurts, human interests. There is the fear of the ‘singularity’, the moment at which AI destroys us. But there are more immediate legal and ethical concerns, such as the use of copyrighted material in training data. ‘Even if pretraining is legal, it may not be ethical.’ He gives he example of the ability of AI image generators to create art in the style of named artists. ‘Why pay an artist for their time and talent when AI can do something similar for free in seconds?’

Then there’s the question of bias. ‘AIs seem to have a generally liberal, Western, pro-capitalist worldview, as the AI learns to avoid making statements that would attract controversy to its creators, who are generally liberal, Western capitalists.’

Mollick’s approach to the various problems and challenges is to propose four principles for working with AI: Always invite AI to the table (you need to experiment since there’s no instruction manual); Be the human in the loop (hallucination is a serious problem); Treat AI like a person (but tell it what kind of person it is) (‘imagine your AI collaborator as an infinitely fast intern, eager to please but prone to bending the truth’); Assume this is the worst AI you will ever use (we are only at the beginning).

The book then presents five perspectives or personas for exploring AI; viewed as a person, as a creative, as a coworker, as a tutor and as a coach. ‘The point here is that AI can assume different personas rapidly and easily.’

This leads to important questions. ‘If AI is already a better writer than most people, and more creative than most people, what does this mean for the future of creative work?’

Mollick is in no doubt that AI will affect your work (unless you’re a dancer, an athlete, a pile drive operator, roofer or motor mechanic).

While AI can automate part of our work, ‘getting rid of some tasks doesn’t mean the job disappears. In the same way, power tools didn’t eliminate carpenters but made them more efficient, and spreadsheets let accountants work faster but did not eliminate accountants.’

To test this, Mollick collaborated with a team of academics and Boston Consulting Group to compare the work of one randomised group of management consultants against another group working with ChatGPT-4 in a series of creative and persuasive tasks (very similar to the tasks performed by public relations consultants, please note).

The conclusion was clear. ‘The group working with the AI did significantly better than the consultants who were not… The AI-powered consultants were faster, and their work was considered more creative, better written and more analytical than that of their peers.’

Yet that isn’t the final word. Boston Consulting Group created one more task that would be difficult for the AI to accomplish. In this task, the human consultants outperformed those who had begun to rely on AI.

So we need to distinguish between those tasks that AI does well and those that humans are uniquely equipped for (he calls these Just Me tasks). Mollick cites the example of joke telling (unless you’re a fan of dad jokes) but also the writing of his book. ‘AI is good at writing, but not that good at writing with a personal style’ though he suggests this may be a temporary situation.

AI is not yet capable of the self-awareness the author shows when admitting ‘I am an academic, and I have the typical academic’s curse of wanting to add too much complexity and nuance to everything.’

To his and the publisher’s credit this book could not be more readable: the meaning is always clear and the font and line spacing are large enough to aid legibility. He even shows how he prompted AI to improve the clarity of his text. ‘Using AI as a co-intelligence, as I did while writing, is where AI is the most valuable.’

Indeed, the section about work (‘AI as a coworker’) is essential reading and will provide inspiration and provoke questions for every professional.

The book can have no conclusion. Instead, the author provides four possible scenarios for the future of artificial intelligence (there’s peril and potential in all of them). His final sentence offers hope, of sorts. ‘Humans are far from obsolete, at least for now.’