• Mark Palmer, General Manager of Analytics, Data Science & Data Virtualisation at TIBCO


• 06.08.2021 09:15 am
#data

Quaesitum means “the true value of a quantity based on measurement.” Evidence shows the way forward… or so the theory goes. But good empirical data doesn’t create miracles. Decisions are often based on emotion, judgement, and experience; empirical data lights the way, like a flashlight in the woods.

That’s right: empirical measurement isn’t the whole truth and nothing but the truth, so help us God. So, what is? A new way of thinking – decision-driven-thinking fuses human insight, intuition, empathy, and nuance, and puts data in its rightful place.

Flying blind in the data-driven vortex

Pablo Picasso said, “The problem with computers is all they can do is provide answers.” Indeed, the art of decision-making is knowing what questions to ask. What Picasso might have meant was that unless we pursue deeper questions, then we may as well be flying blind in a fog of data particles. Computers just make the particles accelerate faster.

Yet somehow, data is now seen as a magic elixir – a higher knowledge.

It’s not. Indeed, data is a core construction material for digital business. But decision-driven thinking turns the current data-driven craze on its head. Where data-driven teams start with technology like data ingestion, deduplication, modeling, and analytics, decision teams start by designing questions.

Questions and data, straw, and drink

Decision-driven is about the unknown, not the known. Questions can be asked of anything: customers, supply chains, macro-economic trends, sustainability, security. When decision teams ask good questions, magic happens. Questions become the straw that stirs the drink, data is an essential element of the recipe. If you have one without the other, the result tastes bad.

For example, consider a home maintenance service. Data-driven thinking starts with data about HVAC service requests, cyclicality, and customer response to offers and promotions. Decision-driven thinking aims to understand the emotion behind customer experience, attitudes towards eco-friendly products and how to talk about them, and what makes customers smile when they interact with the company. For many businesses, these decision-driven questions are a mystery, and they change.

Decision-driven creates a higher tier of intelligence

Using data to explore the unknown is a higher tier of business intelligence. That, though, has been the case since the starship enterprise captured minds in the 1960s. What’s new now?

Artificial Intelligence, as we all know, is in our future. But don’t listen to the sci-fi fanatics: AI doesn’t give better answers, its power leads in helping ask better questions.

The power of deep learning algorithms, for example, is to discover patterns in millions or billions of points of data that the human eye can not see. But note well: patterns only lead to questions, not answers.

Machine learning models are incredible, real-time pattern-matchers. Feed an image recognition algorithm a million images a day, and it will filter the wheat from the chaff correctly 999,990 times. It’s the 10 needles in the haystack that, when passed to a human, help man and machine work together as one.

Enter ModelOps

The last mile between algorithm and human is ModelOps, or the democratisation of how to connect AI decision models into functional workflows, and to humans. In simpler terms, ModelOps helps deploy AI models to use, which drives business decisions. By making it easier to put algorithms into the hands of humans, we scale the higher-level intelligence described above.

A leading analyst firm clarifies ModelOps more broadly at the operationalisation of AI, including decision models, machine learning, knowledge graphs, rules, optimisations, plus linguistic, and agent-based models. Right. That’s complicated, and that’s the other reason ModelOps is an essential straw to stir the complex AI drink.

A new partnership

AI can help us overcome our inherently human weakness in terms of breadth, scope, and cadence of the information and data we are able to ingest and process. But equally, human intelligence and decision-first thinking will help us pose better and better questions that leverage their astounding computational power. 

We need computers, but computers need us, too. Picasso was right – computers aren’t of much value if all they do is give us answers. But remember the order: it’s decisions, decisions, data… and then decisions – with teamwork all along the way.

Further reading:

https://techno-sapien.com/blog/decision-driven-qa

https://towardsdatascience.com/be-decision-driven-not-data-driven-d12b9b7edd8b