What is AI, Really? Part II

In the first part of this module (https://aimind-ft.com/article/what-is-ai-really-part-I/), we defined what is and isn't AI. Now let's look at how AI learns and is implemented in organisational processes.

How does AI actually learn?

To help you understand how AI learn? Let’s take the example of a young child. In order to speak, a child mostly listens sounds, words, human behaviour, gesture… before he would have been able to fully speak as normal individual

For an AI, it is exactly the same: it analyses data which it has access to before producing an answer. To identify if a cat is present in a picture. AI will analyse multiple pictures before being able to identify one.

To perform, AI uses three tools at its disposable to fully filled its tasks:

• Data is the fuel for AI. AI is impossible without data.
• Algorithms are like recipes, giving AI the ability to react to specific situations.
• Computing power is the AI’s engine. The more computing power there is, the greater AI's capability to act quickly.

AI in organisations

In the world of business, enterprises can leverage AI’s power by using for different purposes such as the automation of repetitive tasks. In doing, workers can focus on more valuable tasks like customer relationship or automate the writing of reports to enhance human interaction.

From an engineering perspective, AI can be used for predicting analysis to anticipate clients needs and challenges and give them the most appropriate answer.

At last, AI is also a tool for supporting the process of decision-making. It does that thanks to its ability to analyse quickly large amounts of data using advanced analytics mostly.

However, AI comes with serious limitations and must be used with caution. One of the most significant issues is bias. Since AI systems are designed and trained by humans, they can unintentionally reflect human perspectives and assumptions. As a result, these biases can be embedded in the algorithms, influencing how the AI interprets data and makes decisions.

As mentioned above, AI is highly dependent on data. This means that, if inaccurate or harmful data is fed into a system, the AI may not behave as intended. In addition, AI doesn’t understand context.

A well-known example is Tay, a chatbot released by Microsoft. It was designed to interact with users online, but after being exposed to offensive and manipulative input from internet trolls, it quickly began generating inappropriate and harmful content. This illustrates how AI systems can adopt undesirable behaviours when trained on biased or unfiltered data.

Key Takeaways

1. Data is the fuel quality matters AI is only as good as the data it's trained on. Poor data produces poor AI.

2. Within organisations, AI is used to automate repetitive tasks or to support the process of decision-making

3. AI has limits AI can make mistakes, reproduce biases, and doesn't understand context the way humans do. Knowing its limits is just as important as knowing its capabilities.