How decision intelligence can use data to help create a better world

In the new book, ‘LINK: How Decision Intelligence Connects Data, Actions, and Outcomes for a Better World,’ author Lorien Pratt says we need to take AI to the next level to solve critical social problems.

We all know that we have reached great heights in terms of what science and technology can do for us. But why aren't we using technology to solve the world's truly pressing problems like hunger, poverty, conflict, inequality, unemployment and disease?

In the new book “LINK: How Decision Intelligence Connects Data, Actions, and Outcomes for a Better World” (Emerald Publishing; September 16, 2019) author Lorien Pratt explores just how we can use technology to make a better world for us. In doing so, Pratt says we will need to take artificial intelligence to the next level and use decision intelligence (DI) in areas like economics, optimization, big data, analytics, psychology, simulation, game theory and more.

DI is a new field with inevitable momentum. It is the next “layer” on top of AI, helping this powerful technology to solve the most important problems we face, Pratt explains. Both Google and SAP have decision intelligence organizations now, and DI in active use for planetary defense at NASA’s Frontier Development Lab. As Pratt explains, DI is also an important force in understanding the societal impacts of AI.

What exactly do you mean by the term decision intelligence?
Lorien Pratt: Decision ontelligence is the discipline of understanding how actions lead to outcomes. Because a "decision" is what we think about when we are considering taking an action. The discipline is the "glue" between many other fields, integrating artificial intelligence, economics and much more.

What are the top themes that you address in the book?
Pratt: Several:

1. What is decision intelligence and how did it arise to fill a gap between technology and humans?

2. Why is it important that technology be democratized: made accessible to many more people than use it today?

3. Who is doing DI today? About the companies that are pioneering DI, including Quantellia, Google, Alibaba, and more.

4. How can we draw a map of a decision (called a causal decision diagram)?

5. Why hasn't AI realized its full promise, and how was DI designed to fill this gap?

6. How does DI connect to other technologies and fields, like systems analysis, decision analysis, economics, and more?

7. What do climate change, poverty, conflict, and other large and complex problems have in common, and why have they not been solved today and how does DI help?

8. How does DI solve the problems that have arisen from silo-ization within complex organizations?

How does decision intelligence differ from the other technology areas in which you have worked – artificial intelligence and machine learning?

Pratt: ML handles single links, answers questions, and provides insights, classifications, and regression. DI combines single links into a causal chain from actions to outcomes. Because of this it bridges the gap from the data/AI stack to end users.

A lot of non-ML AI solutions focus on building knowledge bases, but these are, by and large, "boil the ocean" efforts without a tight focus on the topics of interest to decision makers in companies and governments. DI creates a tight focus on knowledge, because it focuses on the specific knowledge elements needed to decide what action to take to achieve some outcome (e.g. revenue, societal outcomes, and more).

ML today focuses, by and large, on fully autonomous use cases. These are a tiny subset of all use cases. DI brings humans, along with their expertise about how systems work, into the loop in a formal and systematic way.

You advocate using decision intelligence to take AI to the “next level” to help improve conditions in the world. Can you provide some examples of what you envision here?

Pratt: Certainly. They include:
  • Create a decision model to select from several legal interventions based on mapping the interaction of belief in the rule of law, a sense of security amongst people, the strength of the legal system, and more (this was the prototype work we did for Liberia).
  • Instead of just using ML to predict customer churn, surround the ML model with a decision (action-to-outcome) model that helps to decide which churn reduction method will be the most effective with each customer, in order to maximize this year's and next year's net revenues.
  • Instead of just using ML to detect a security intrusion, surround it with a decision model to decide which action to take for which kinds of intrusions that maximizes net revenues (because different security treatments have different costs and outcomes)
  • Instead of just using ML to identify where a climate-change-driven flood is likely to impact a country, use DI to choose between several prevention / remediation measures to minimize the cost of property and lives lost.
  • Instead of just using ML to forecast when a neighborhood will run out of internet bandwidth, surround the ML model with a decision model to prioritize where to expand capacity, taking into account competitors in the area, expected subscriber growth, and upcoming pricing plans, and to calculate customer satisfaction, in order to maximize net revenues this year and also 5 years from now.
  • Instead of just using ML to analyze the movement of asteroids, surround the ML model with a DI model to decide what to do to deflect on that is aiming for Earth (this is the's "deflector selector").

At we at the point now that we can use decision intelligence to address these issues, or does more technological advancement need to happen first?

Pratt: Yes, we are ready. You can learn useful DI in five minutes, and there is a series of increasingly more sophisticated technological approaches that can be used as are justified.

In addition, there are many opportunities for advancing DI technology. It's an exciting, and wide-open, field.

Are these issues that governments must address using advanced technologies, or corporations and non-profits, or a combination?

Pratt: Because of their duty to ensure equality and safety for all people, governments are particularly good locations for DI, so as to map and surface unintended consequences from policy decisons that might play out over several years.

It’s not all good news when it comes to the public at large embracing AI and other advanced technologies. There is a great deal of fear as well. Should the public be cautious about decision intelligence?

Pratt: All powerful technologies can be used for good or for bad, and decision intelligence is no exception. However, given that the nature of DI is to shine a bright light on unintended consequences, hidden agendas, and vicious cycles that are otherwise hard to understand in complex situations, DI has a particular role to play in helping to solve the problems in AI and other technologies.

Indeed, I co-organize the annual Responsible AI/DI Summit ( ) where influential thought leaders meet annually to explore ways to maximize the positive impacts of these powerful technologies while controlling the downsides.

Any parting advice or observations on the topic you would like to share?

Pratt: DI is an inevitable next step in the advancement of the human/computer partnership. I invite your readers to learn about this important new technology, to join my ecosystem list by subscribing to my blog, and to check out

More for you

Loading data for hdm_tax_topic #better-outcomes...