Co-Intelligence summary

To fully utilize the potential of AI, it’s essential to have a baseline understanding of what it is. There are many different forms of artificial intelligence – including the types that can generate images and videos like deep-fakes. But this Blink will focus specifically on one type of AI called large language models – or LLMs. This is the AI that generates language, and is set to enter many of our workplaces in the near future.

LLMs work by analyzing the language patterns found in massive datasets made up of written materials. Through this analysis, AI can then predict a credible response to a question or request that a human might ask it. And because the datasets used to train AI have been written by humans – they are essentially made up of books, papers, and web content – the responses have a compellingly “human” tone to them. This means that when you’re using AI software, you’ll feel like you’re working with another human, rather than a computer or a robot.

So, exactly what kind of personality does AI have?

Well, there are a few qualities that are central to all LLM-based AIs.

First, they’ll consistently surprise you. This is a core difference between the latest generation of AI and its predecessors. Keep in mind that AI is a type of software. Historically, software was designed to reliably adhere to a prescribed pattern, in order to achieve a specific end result. But LLMs don’t work like this. They’re designed to be generative – finding solutions to complex problems and delivering creative outputs. This is one of their core benefits.

However, there’s a major disadvantage to this quality as well. Generative AI is inclined to tell lies, which are known as hallucinations.

An LLM-based AI can’t distinguish between fact and fabrication – it’s simply programmed to look for patterns and deliver innovative and convincing responses. This means that it’s prone to inventing things and presenting them as truth, even if they’re complete fiction. We’ll delve into this more later but it’s important to keep this quirk in mind as you explore how AI can increase the quality of your work activities.

To make up for this shortcoming, AI possesses an enormously beneficial quality – adaptability. As mentioned earlier, working with AI feels very much like communicating with an actual person. But you, as the human, get to tell the AI what kind of persona you want it to have each time you use it. For instance, if you want an LLM to prepare a business report, you could say to it, “You are a highly intelligent and astute business manager who is articulate, persuasive and transparent. Write the executive summary of a business report, based on the input document.” The output your AI delivers will then be in an appropriate tone. Alternatively, you could tell the AI to assume the persona of a hopeless romantic, and ask it to write a poem about sunsets for your beloved.

Essentially, working with AI means that you have an endless array of personalities and experts at your fingertips, all ready and able to present you with what you need, in the style that you need it. This doesn’t mean that you’ll be able to fully outsource every language-based task to AI – we’ll explore why in the sections ahead – but it does mean that you can learn how to best exploit AI to increase your productivity and performance. In the upcoming sections, we’ll look at how you can do this in various work scenarios.