What are problems with AI (artificial intelligence)? What we should know about AI’s troubles?
When we teach a computer to learn how to respond, we use fixed parameters. The AI must recognise these parameters before it will respond. The input data must not be too similar or else the machine has very limited terms and conditions to recognise. If and when it encounters non-conforming terms and conditions, this machine has no precedent to refer to, to understand what response it has to show.
How do the scientists deal with this problem? The solution is to teach the computer processor to recognise a wide variety of input stimuli, and yet choose the appropriate response to handle the situation. While there is a preferred method to deal with a specific set of circumstances, the machine has to learn to ignore extra confusing data and focus on the essential parameters. While concentration for correct orientation is good, this behavior results in narrow mindedness.
Now imagine what can happen when you ask an AI to pick a profitable stock for you to purchase. The artificial intelligence will run a set of factors through its pre-determined algorithm.
- Has the stock performed well in the recent past?
2. Is it sensitive to world economy?
3. What are the raw materials and supporting networks that can impact its price?
4. What are the current factors that influence the prices of the materials and networks?
5. Is it more dependent on domestic economy?
and etc.
The AI can think “out of its box” only if humans have programmed it to recognise and respond to a wide range of factors, and identify any new input that has impacted the performance of the target.
pigeonhole inputs as
