ARTIFICIAL INTELLIGENCE: WHERE DOES ONE BEGIN?
If you were to look over my shoulder at my news app, it wouldn’t be, beyond the 2nd or 3rd article at best, that youend up stumbling upon AI or ML in some form. Almost every startup and tech company today claim to use AI in their product or service to give you a better experience, value and of course the coveted three letters of RoI. In spite of all this, we aren’t really seeing ‘a revolution’ that was promised to us and some studies even show that AI research is hitting a sort of a ceiling. However, there is a quieter revolution that is occurring, even though it may not be too apparent at first. Let’s take a look at Google. They’ve built a formidable tech business over the past couple of decades and are eyeing for the prized trillion-dollar valuation today.
Google redefined their mission form ‘wanting to organize the world’s information’ to being an ‘AI company’. That’s huge. Companies the sizes of Google don’t go about ‘pivoting’ their mission. And yet, it seemed the most obvious thing to do for them. So, what does that mean? How has it played out? If you pick Google Assistant, it’s way ahead of the other ‘assistants’ in the market. Alexa, Siri (Bixby anyone?) are left far behind in the conversation prowess that Google Home (which is powered by the same Google Assistant technology) possesses today. While Alexa is still leading from a hardware sales stand point due to their early start in the US market, Google Assistant is already used by far more users due to the penetration that Android has. Voice search has been growing in double digits every year in almost all markets for the past 5 years and its growth doesn’t seem to be slowing down.
Take the same with Google Maps. Its accuracy to estimate traffic and routes is something the companies like Uber have built their businesses on. Imagine how Uber and Ola would be able to manage their fleet and give you a seamless point-to-point experience without Maps? Navigating and estimating Maps with old statistical methods is just not possible and AI is the only way to achieve this level of accuracy and scale.
All this talk on AI gets one to think, how do I bring AI into my own organization? How do I begin to think and act like Google. It’s here where many organizations begin to struggle. Most organizations get mislead when they are evaluating big data and machine learning. They begin with evaluating and subscribing to some magic bullet AI product or get swayed away by Data Consultants. But, that’s not what is needed.For any AI to work it needs data. And it needs lots of it. Loads of data that is prepared in a form that the AI can work with. When organizations realize this partially, they again seem to flounder. I’ve seen IT teams go for Hadoop trainings and planning to build a data lake. Talking about command centers and build castles in the air, before even understanding the ground reality of what is needed.
The first step is to get a buy-in from every stakeholder in your organization that Data is the life blood of the organization and it needs to be accessible and flowing in the right manner across the various stake holders and functions to facilitate the right decisions, actions and their feedback. This is harder than you think. The way in which organizations are run is by having ‘departments’. These departments, almost always work in silos. They have ‘their own KPIs’ and ‘their own language’. This makes it difficult for them to talk to each other. Data makes it possible to do it, however, that is not how they were built. I’ve personally seen cases where such discussion takes not just months, but years. This is where an ODS or a ‘One Data Source’ is crucial for any organization that is leading towards the data path.
The second step again is not to go after some software, but to do a proper data audit. To understand what all data you have accrued over the years, to be able to assign confidence levels to different data sources, to identify which need to be upgraded and which new data sources could we begin to tap on. This might mean collecting your customer’s phone numbers via a sampling program. This being done on a tablet device with OTP verification assures your data quality is high and not lost due to bad handwriting or intent. This is again a very critical step. Without good quality data, the best of algorithms will fail.
The third and final step is to establish a data driven culture. You could build the best databases, have the best processes for collection, the best engineers to analyze the data, but if you don’t have a data driven culture, then egos and biases will reign supreme in your organization. All in all, if you ask someone at Google or Facebook, they will tell you, that they are not AI or social media companies, rather they are ‘data’ organizations. If you want to succeed, you need to be a data organization no matter what industry you belong, what you are currently selling or to whom you are selling to. And it doesn’t start with any technology, but like all wise things in life: It begins with you!