Page 144 - Computer - 7

improved by direct human intervention. This makes symbolic AI less effective for complex problems
where not only the variables change in real-time, but also the rules.

Millions of ‘if-then-else’ rules could not capture all of a doctor’s domain knowledge and expertise, nor
their continual development over time. Despite these limitations, symbolic AI remains far from obsolete.
It is particularly useful in supporting humans working on repetitive problems in well-defined domains
including machine control and decision support systems. The reliable performance of symbolic AI in
these domains has earned it the endearing nickname ‘good old-fashioned AI’.

MACHINE LEARNING AI

Machine learning (ML) refers to a wide range of techniques which automate the learning process of
algorithms. This approach differs from the earlier approaches whereby improvements in performance
are only achieved by humans adjusting or adding to the expertise which is coded directly into the
algorithm. In ML, the algorithm usually improves by training itself on data. This is why Machine Learning
AI is also called data-driven AI.

Rather than having their knowledge provided by humans in the form of hand-programmed rules,
Machine Learning systems generate their own rules. For Machine Learning systems, humans provide
the system training data. By running a human-generated algorithm on the training dataset, the Machine
Learning system generates the rules such that it can receive input x and provide correct output y.

In other words, the system learns from examples, rather than being explicitly programmed. This is why
data is so vital in the context of AI. Data is the main raw material out of which high-performing Machine
Learning AI systems are built. The quality, quantity, representation, and diversity of data will directly
impact the operational performance of the ML system. Most ML systems run on general-purpose
computing hardware, and nearly all of the best algorithms are freely available worldwide. Hence, having

the right data tends to be the key. While it is true that Machine Learning systems program themselves,
humans are still critical in guiding this learning process. Humans choose algorithms and datasets, format
data, set learning parameters, and troubleshoot problems.

142