Sunday, August 31, 2025

Why is it so hard to order Tacos from AI?

   Why is it so hard to order Tacos from AI?



When we see cases where companies like Taco Bell fail in their plans to replace human workers with AI, we might ask what went wrong. The answer is simple: the AI didn’t understand when somebody asked something. And then somebody asks, Why does AI understand better when a professor or military leader says something than in other cases. Where does a regular person ask something? The answer is that those first two cases speak using literary language. Regular people speak using dialects. And that is hard for the AI.
When AI translates speech, it uses a speech-to-text application. This application transforms spoken words into a form. That algorithms understand. “Understanding” means that the system searches the algorithm. There is a match for those words or commands. That match is like a trigger that activates an action. That the algorithm controls.
If a person uses dialects, there must be their own algorithm or trigger for those dialect words. This means that the fuzzy logic that the AI uses. Is the sum of the precise logic. The fuzzy logic means that there are multiple triggers that the system can activate. So every dialect requires its own trigger. Or in other cases, there must be a translation protocol that transforms dialect words into literal words.
And then that system transmits the translated words to the trigger that activates the action. The AI has problems in cases where somebody has speech defects. Or they use something else. Than the official dialects. In those cases, there is no match in triggers. This is one of the biggest problems. For large language models. LLM.
When the AI takes command, it follows a certain line. First, it must search keywords like “draw an elephant for me”. The system finds the word “draw”, which means. The system transfers the mission to an algorithm or module that handles graphics. Then the system searches for images. Where there are elephants. The system creates its own version of those images. It connects those images using certain parameters. And that means the image that AI creates is the average of the images that the AI uses. The AI must understand the words that it gets.

https://futurism.com/taco-bells-ai-drive-thru

No comments:

Post a Comment

Note: Only a member of this blog may post a comment.

Controlling qubits on the next level.

“Caltech has built the world’s largest neutral-atom qubit array—6,100 qubits—pushing quantum computers closer to error correction and full-s...