Infinite Analytics

Stories

The case for and against democratizing artificial intelligence

by Team IA, January 27, 2021

Democratizing artificial intelligence means making it accessible to all. This involves providing access to AI tools and data required to build AI systems. Democratizing AI will thus lead to better innovations, more efficient AI systems, and even better engagement. In the current state of affairs, cash-heavy big firms and tech-savvy start-ups are the major users of AI.

The aim of democratizing AI is to make it available across a broad spectrum of entities such as government agencies as was demonstrated during contact tracing for Covid and prioritizing groups for Covid vaccine distribution, and even extending it to smaller entities or even individuals. A good example of democratization of AI is that of Makoto Koike of Japan. In 2015, Koike, while helping his farmer parents to sort cucumbers at their farm, used Google’s Tensorflow, an open source library. A ‘deep-learning’ algorithm was able to help him classify cucumbers based on photos. The results were far from quite accurate due to the small dataset. The images fed to the system were also of a low-resolution. But his story went viral on Youtube as it told a tale of the hitherto unharnessed potential of AI use across human tasks.

The problem with democratizing, however, arises due to the lack of knowhow on coding, statistics, data analysis etc., simply due to an inadequate number of personnel equipped with such knowledge. Even if they are, the cost of hiring them is exorbitant.  So, tech giants such as Google, Amazon et al offer easy-to-use cloud technology and easy-to-use AI tools such as virtual assistants and even Natural Language Processing, which can be easily incorporated into apps sans any knowledge of making machine learning (ML) models. While these are used to boost businesses, such democratization, though affordable, can be dangerous and highly susceptible to bias.

Cheaper cloud tech also means using the services of newly minted AI specialists with less experience and knowhow, leading to use of such data which is not advanced in its quality. The use of poor quality data may unleash negative effects on the firm or user, and may not be detected until the point-of-no-return is reached. This is where the role of specialized data scientists becomes imperative to oversee smooth functioning of AI.

It is also critical to ensure that people who use such user-friendly AI tools should have some knowledge about the functionality of such tools and not simply the know-how to operate them.

Simply extending AI tools without in-depth knowledge equals the most common form of untrained danger such as asking a dental surgeon to operate on a limb. Similarly in the field of AI, the lack of interpreting data in the correct manner can be just as hazardous.

Another common issue faced while implementing AI democratically is the limitation of its knowledge to the top leadership. Unless employees at the bottom of the ladder also know how to use AI, data transparency cannot be achieved. A fitting example is that of Airbnb, who are implementing data transparency by giving access to employees and thus helping them in decision-making.

Data is now available to everyone and we at Infinite Analytics are committed to make gold out of all your data. Our platforms and tools are home made and global. Moreover, Democratizing AI cannot happen in an instant. It will work only with in-depth and wider knowledge and commitment to sharing the complete knowhow instead of just the upper realms of data-science. You can follow our posts and insights at Infinite Analytics (substack.com)

Subscribe to our newsletter

* indicates required



   


Avatar

by Team IA

Leave a comment

Your email address will not be published. Required fields are marked *