Can Artificial Intelligence Lie?

Just finished reading a very interesting blog by John Harald Oien called “Why AI ‘lie’, and what we can learn from it.” (source: Why AI “lie”, and what we can learn from it | Mews Developers).

In this Blog, John points out a lot of things that, for the AI aficionado, is second nature. For instance…

1.       AI models learn through a process called training.  True!

2.       Supervised learning is one the most popular approaches to training AI models. Yup!

3.       To store the reams and reams of data required to train an AI model, the AI compresses its model. Not that obvious but it is true… this leads an AI model to, potentially, associate data incorrectly.

Here’s a very important takeaway from John’s blog… compressing billion of data points through a neural net does not necessarily connect the dots in the same manner or with the same output each time. So, arriving at different results is not necessarily a “lie”, just different answers based on what was asked and how it was asked.

So, what’s a lie? Well… lying is intentionally providing false information. Ah! Now we’re getting somewhere.

What does intentional mean? Every dictionary in the World defines intentional as “planned or as intended, not accidental or random.”

For all intents and purposes, an AI model is NOT accidental nor is it random. AI models follow programmed instructions and algorithms, even generative AI models do. So, if an AI model “lies”, its programming must have a “lying” component intentionally incorporated. I can dive into a dissertation on AI architectures to prove this point but that’s for another time.

Intention is the driving force behind actions, revealing the purpose and direction of decisions. Conscience, on the other hand, acts as an internal compass, guidance to discern right from wrong and imbuing actions with moral weight. Together, conscience and intention shape the ethical framework within which people navigate their lives. So, now we’re getting to the root of question, “Does AI have a conscience?”

AI doesn't have a conscience or self-awareness. AI processes information and provide responses based on data and algorithms without feelings, beliefs, or a sense of right and wrong. The unique ability to contemplate morality, emotions, and complex social interactions is still in the realm of humans. AI can assist, analyze, and even surprise with its capabilities, but the depth of consciousness remains uniquely human… for now.

That was quite a “Rabbit Hole”, wasn’t it?

So, “Can Artificial Intelligence Lie?” I’d love to hear your answers.