Most up-to-date and relevant AI + LLM courses & coaching
AI vs IT – What Senior IT Professionals need to know before building an AI application
IT is deterministic while AI is probabilistic.
When you make a software application and if it is working properly, the result will always be the same.
For example – If you design a login system and if a user provides the right username and password, the system will always allow the user to log in
However, AI systems are different. AI systems/ models are probabilistic (or stochastic)
Let’s say, you have created an image classification model – cat vs dog. In this case, the model based on its training will figure out the probability of an image being a cat.
And based on this probability, it would predict if it’s a cat or not.
Now, there could be a chance that a model can classify the same image of a cat as “cat” in one instance and “dog” in another.
This is because of AI models’ probabilistic nature and that is the reason, there is no 100% certainty in the outcome.
If you are a Senior IT professional and looking to learn AI in simple language, check out this course – https://www.aimletc.com/online-instructor-led-ai-llm-coaching-for-it-technical-professionals/
AI vs IT – Why does it matter to senior IT folks?
It might seem that AI has many applications and use cases. It’s not exactly the case.
For example – A big use case of AI systems could be in “cancer detection systems”. However, it is still not that popular, why?
Because of the AI system’s probabilistic nature, in such systems, even a slight inaccuracy is not acceptable. Both the cases are dangerous –
- When the AI system detects cancer when it’s not, and
- AI system does not detect cancer when it is present.
And because of this inaccuracy, AI systems are not entirely helpful.
So, where does the AI system help? Where the accuracy is not a problem.
For example – recommendation systems. Whether it is YouTube, Instagram or Netflix, we receive AI-based recommendations.
Here the system works because both scenarios do not matter much, whether
- It shows you the video that you do not like to watch, or
- It does not show the video that you like to watch
Similarly, other use cases of AI that are popular are WhatsApp’s autocorrect or Gmail’s autocomplete feature.
In both the features, it does not matter whether the suggestions are 100% relevant or not.
If the requirement is such that even a slight inaccuracy is not acceptable then we could not depend on AI.
Either we need to keep a human in the loop and check the output in some way before giving it to the user.
The moral of the story is that AI can only be used where we are fine with its probabilistic nature otherwise, it can not be stand stand-alone application.