"What is AI?". For starter let's assume AI is just a tool to benefit human kind. But human kind is not perfect and humands do mistakes. So if AI aka artificial intelligence continues to mimic human intelligence then will it be free to do the same silly mistakes as the humans do? Actually this thought coins another term the "rationality". Rationality means to do the right thing given what it knows. So this makes two types of AI. One to mimic human capabilities as they are and another one to make only the right choices ignoring human flaws. The second approach is more popular than the first as in our day do day life we all try to make as less mistake as possible. But from time to time the definition of intelligent system changes and after reaching a certain milestone we try to achieve much higher picks. Take for example a program named Eliza. Eliza emulates a Rogerian psychotherapist. He is programmed with a preset of texts and some common text patterns.
Then we know all about the story of Garry Kasparov and deep blue. World chess grand master Garry Kasparov was first ever was beat in an intellectual decision making on chess board. In 1997, New York City human was defeated by a mere computer program. This was a pretty big event and we officially started to believe machines can be intelligent too.
After these consecutive events AI had to split into specialization sub fields. These fields focus on specific parts of intelligence and its inception. Some fields foucs on human consciousness, some on human intelligent behavior development, solical or mob decision making and so on. Among them a broad and well studied field is Machine Learning, focusing on how dumb systems can learn alike humans and make decisions accordingly. Both academia and industry has embraced this field cordially and currently every sector is adapting to machine learning to get their task process automated intelligently.
Machine learning in industry then gives birth to certain profession and emphasis more on data science, the science that works with data to extract and use as many information as possible. Lets look into a few,
- Data engineer: Works for structuring and managing data. He is responsible to maintain and design data warehouse. But generally he is not responsible for getting insights from data. Managing distributed data source and their accessibility is another important duty for him. Generally companies that have massive historical data start with these professionals. They actually make way for data analysis as it is almost impossible to make sense of data from massive unstructured sources.
- Data analyst: Data analyst analyse data and finds insights out of them. From time to time they use well established machine leanrning algorithms and help to make important and business critical decisions. Finding out revenue points and cost cutting places of a department is some of his job. Also finding the right machine learning algorithm for prediction is his missioin critical jobs.
- Data scientist: Data scientists are the most critital jobs in this sector. Most of the time new machine learning algorithms are build by him. He analyze the situations and tries to reach a generic point where he can propose an algorithm to solve it for all those cases. His work is a long term asset for an organization.
But this hierarchy is for large well established organizations. Machine learning is more of a culture for an organization. To start this culture we may start using the existing services. Then extend those with a coherent team and lastly make new breakthrews.
- Artificial Intelligence a Modern Approach. Peter Norving, Stuart Russell
- Eliza: http://psych.fullerton.edu/mbirnbaum/psych101/Eliza.htm