Dr Mike Hobbs, Visiting Researcher in CEL and an expert in data, AI and AR/VR shares the first is series of short postings:
These days, if we are involved in research, policy or just trying to find evidence for good practice we need data science.
Data Science Central is a site run by, and for people, who are professionally engaged in analysing and handling data in the era of big data. A recent tweet, from an experienced practitioner refers to a useful short article that helps to demystify the role of AI in data sciences.
As an academic specialising in data analysis and AI it is refreshing to see AI put into the context of human decision processes and the relevant latest technical advances. The emphasis is on Artificial Neural Nets and machine learning with short notes on the nature of a range of different types.
Interestingly ‘deep learning’ does not get top billing as it is correctly viewed as an adaption of previously existing technology, originally developed in the 1980s. This does not mean that Deep Learning does not provide significant advances on previous technology, but it achieves most of its performance from the benefits of faster processing, larger training data sets and advanced data handling.
Artificial Intelligence, Araujo dos Santos, L, Github
If you think of AI as a steam hammer what has happened is an increase in the number and size of pistons, able to handle more steam with a much fitter stoker hurling coal into the furnace. The data storage and processing from networked cloud computing resources provide a very big hammer for your data analysis problems.
A word of caution – as this article points out, AI is not magic and is based on firm theoretical and statistical principles, which can be abused to create erroneous results. AI typically finds information hidden within large and complex sets of data through various pattern matching techniques but is only as good as the data provided.