Reliable data – a core underpinning of Artificial Intelligence

Artificial Intelligence (AI) is possibly the hottest of hot topics at the moment, particularly when it comes to reliable data. ​

What makes headlines is AI’s combination of wonderful potential applications; everything from curing cancer to driver-less cars, and a potentially terrifying loss of human control in which doomsday scenarios proliferate.

While the upsides seem evident, many people in business are concerned about AI, as they don’t know how it works and in effect, can’t see inside the box.   This then leads to questions of trust and thereby to the ethics of the applications.

Reliable Data

Neural networks, algorithms, the ability of machines to learn are pretty complex and can be esoteric.

That said, one thing that is generally straightforward is the data that is being applied. And it seems just like in every other data-driven dimension of our world, reliable data is the key to successful AI.

At Data Agility we’ve long used the term ‘reliable data’ rather that ‘accurate data’, ‘trustworthy data’ or anything else to describe what data people are seeking to use.  When we spoke with our clients, our research showed that people wanted data they could rely on to make decisions.  Reliability took out absolute standards of accuracy and trustworthiness.

In our day-to-day lives, humans are generally very good at filtering out rubbish data, and the majority of us bring natural statistical skills to our lives which help us through the day.   Similarly, many AI solutions can behave in a comparable way IF there is an understanding of the quality of the data at the outset; that it is reliable data.  If this understanding exists, then training routines can be established that allow the AI identify valid patterns in the data.

From there, the AI can determine what to do, within its preset parameters, with the patterns it observes.   And from there it can continue to learn and deliver useful outcomes.

However, without this baseline data quality understanding there are very real risks of the old fashioned rubbish-in-rubbish-out model being repeated.

So, if you want to implement AI in your business or you are talking to an AI vendor, make sure that you get to the bottom of what data you/they will use and deeply understand the reliability of your data.

0 Comments

Submit a Comment

Your email address will not be published.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

Most popular insights.

Balancing Data Collection and Data Privacy

Balancing Data Collection and Data Privacy

Organisations are collecting more personal data than ever. Whether to provide a better user experience, improve services and operations, innovate, predict trends, or stay competitive, organisations can extract a lot of value from consumer data.

Manage data collection processes

Manage data collection processes

How can organisations effectively manage data collection processes and avoid the risks of data overcollection?
Writer: Rob D’Astolto