In 2023 the professional networking social platform LinkedIn launched a new type of content to encourage more engagement among users. These articles are called ‘collaborative articles’ and are AI (Artificial Intelligence)-generated. They are accompanied by an invitation or prompt at the end of each paragraph to contribute thoughts and opinions in the comments.

This phenomenon caused me to start brainstorming ideas about the current opinion piece on the difference between what machines learn through artificial intelligence and what humans learn through knowledge.

Let’s take a journey exploring the evolution of AI and how it differs from human knowledge.

Disclaimer: no AI tool was used to write this article. Humble brag: contact me for a content writing quote.

A Quick Word on LinkedIn’s Collaborative Articles

To acknowledge the fact that LinkedIn gave me the inspiration to write this article (in my personal opinion, LinkedIn is ‘the watching paint dry social network’), let’s delve briefly into their concept of ‘collaborative articles’. In a nutshell, these articles are generated by AI using information that is readily available, they often contain sweeping generalisations and simple language to explain a concept without quoting any references. This is a particularly worrying element of these articles as there is no mention of sources, making them liable of being a product of so-called “hallucinations” or, simply put, statements that are not backed by evidence.

Then, these articles are pushed to users’ timelines with a request asking for comments. LinkedIn users that have the required skills and expertise can add a comment to each section of an article, so that the general public can “learn from experts across professional topics”. In other words, these experts are unpaid to contribute to an article that amounts to zero cost to LinkedIn in exchange of ‘visibility’ and to be perceived as knowledgeable in their field.

The so-called ‘conversation starters’ created by AI following prompts by LinkedIn’s editorial team require the input from experts to have any real value. On their own they are simply mini-essays that wouldn’t be worth reading. This is a reminder that AI offers tools that could help with content creation but they are not a replacement for it and that intelligence is the ability to use acquired knowledge in an efficient way (source).

What Is the Difference between AI and Human Knowledge?

Let’s start from the basics. The definition of knowledge is: “acquaintance with facts, truths, or principles, as from study or investigation” (source). The definition of intelligence is: “the ability to think, reason, and understand instead of doing things automatically or by instinct” (source). The definition of Artificial Intelligence is: “the use of computer programs that have some of the qualities of the human mind, such as the ability to understand language, recognize pictures, and learn from experience” (source).

What does this tell us? It tell us that AI, rather than being an exact mirror image of human knowledge, mimicks some characteristics that we usually associate with a set of human activities. AI requires an existing amount of data to interpolate into a resulting response, be it an image file, text, audio or video. The key word here is “existing”: while we commonly talk about generative AI implying it has creative qualities, AI is derivative in its output.

Does AI Have Knowledge?

Knowledge is one of the main shortcomings of AI.

What is common knowledge in AI?

When it comes to artificial intelligence research, it should include what humans perceive as being commonsense such as shared observed facts about everyday life, for example the grass is green and the sky is blue. These are concepts that humans learn from an early age and that are taken from granted. However, AI has to be told all of these concepts and they should be somehow protected from being altered, just like a Wikipedia page gets locked from being changed by unscrupulous editors.

What Can We Learn from This?

Ultimately, human input is what will make or break AI and the quality of its output. There is an increasing demand for human intervention to run quality assessments and fact-checking of AI-generated content.