Bank of England’s Head of Data Collection talks data integrity
In this week’s episode, Seth Adler talks to the Bank of England’s Head of Data Collection & Publication, Beju Shah. Here is a sneak preview of how the discussion unfolded on why the success of sophisticated technologies like artificial intelligence (AI) depend on clean, organised data that follows a framework of standards.
Beju was involved in building systems that featured in studies on improving eyecare standards in the UK. He notes that these projects taught him: “...before anything else, you must get your data correct. [That] is fundamental before you get to the applications and really start applying AI at scale. This includes clear identification, clean data and standards for structured data. If you build up that good foundation, you have the necessary qualities of data to input into more intelligent processes. If the data going into intelligent processes isn’t very good - it’s not going to be very intelligent, is it?” In other words, GIGO.
Beju himself, quotes Seth Goddan; “We often forget to spend time and attention on the preparation that is a lot less urgent or glamorous - but [is] far more important.”
This clean data unlocks the potential of artificial intelligence. AI gives “people new ways [of looking] at data, but more importantly it takes out all the manual work of quality checking. [It allows for more] meaningful work, and [lets] your talent shine - that’s what the technology is there for - to help us shine and bring our talents up… It allows us to do our best and most meaningful work.”
Data integrity projects are not an overnight process. Rather than waiting for the perfect data lake, Beju notes that businesses should experiment to learn how techniques such as machine learning or AI data inserts could be applied in certain localised areas. “It’s [about] looking at how we’re getting insights from unstructured data so we don’t have to trawl through lots of things and do all this manual work.”
To those of you out there pumping money and resources into improving your data lakes, tech-stacks and skill-sets, is there a small part of you that fears your business needs will outgrow these great foundations in say 15 years because of the speeds technology and customer requirements are developing at? Fortunately, Beju believes this road-block is becoming a thing of the past. He says: “ What we’re doing now is using web technologies that anyone can integrate into any system. Whatever your legacy, whatever your start-up is, or how advanced you are, these open standards allow for the flexibility to just do it. Modern tech has evolved and matured.”
Find out more about Beju, his background and what he thinks the future of tech looks like by clicking on the podcast below!