The next BriefingsDirect business intelligence (BI) trends discussion explores the latest research and products that bring the power of people and machine intelligence closer together.

As more data becomes available to support augmented intelligence — and the power of analytics platforms increasingly goes to where the data is — the next stage of value is in how people can interact with the results.

Stay with us now as we examine the latest strategies for not only visualizing data-driven insights but making them conversational and even presented through a form of storytelling.

<br />

Listen to the podcast. Find it on iTunes. Read a full transcript or download a copy.

To learn more about making the consumption and refinement of analytics delivery an interactive exploit open to more types of users, we welcome Elif Tutuk, Head of Research at Qlik. The interview is conducted by Dana Gardner, Principal Analyst at Interarbor Solutions.

Here are some excerpts:

Gardner: Strides have been made in recent years for better accessing data and making it available to analytics platforms, but the democratization of the results and making insights consumable by more people is just beginning. What are the top technical and human interaction developments that will broaden the way that people interact differently with analytics?

Tutuk: That’s a great question. We are doing a lot of research in this area in terms of creating new user experiences where we can bring about more data literacy and help improve people’s understanding of reading, analyzing, and arguing with the data.

In terms of the user experience, a conversational aspect has a big impact. But we also believe that it’s not only through the conversation, especially when you want to understand data. The visual exploration part should also be there. We are creating experiences that combine the unique nature, language, and visual exploration capabilities of a human. We think that it is the key to building a good collaboration between the human and the machine.

Gardner: As a result, are we able to increase the number and types of people impacted by data by going directly to them — rather than through a data scientist or an IT department? How are the interaction elements broadening this to a wider clientele?

Tutuk: The idea is to make analysis available from C-level users to the business end users.

If you want to broaden the use of analytics and lower the barrier, you also need to make sure that the data machines and the system are governed and trusted.

Our enterprise data management strategy therefore becomes important for our Cognitive Engine technology. We are combining those two so that the machines use a governed data source to provide trusted information.

Gardner: What strikes me as quite new now is more interaction between human cognition and augmented intelligence. It’s almost a dance. It creates new types of insights, and new and interesting things can happen.

How do you attain the right balance in the interactions between human cognition and AI?

Tutuk: It is about creating experiences between what the human is good at — perception, awareness, and ultimately decision-making — and what the machine technology is good at, such as running algorithms on large amounts of data.

As the machine serves insights to the user, it needs to first create trust about what data is used and the context around it. Without the context you cannot really take that insight and make an action on it. And this is where the human part comes in, because as humans you have the intuition and the business knowledge to understand the context of the insight. Then you can explore it further by being augmented. Our vision is for making decisions by leveraging that [machine-generated] insight.

Gardner: In addition to the interactions, we are hearing about the notion of storytelling. How does that play a role in ways that people get better analytics outcomes?

Storytelling insights support

Tutuk: We have been doing a lot of research and thinking in this area because today, in the analytics market, AI is becoming robust. These technologies are developing very well. But the challenge is that most of the technologies provide results like a black box. As a user, you don’t know why the machine is making a suggestion and insight. And that creates a big trust issue.

To have greater adoption of the AI results, you need to create an experience that builds trust, and that is why we are looking at one of the most effective and timeless forms of communication that humans use, which is storytelling.

To have greater adoption of the AI results, you need to create an experience that builds trust, and that is why we are looking at one of the most effective and timeless forms of communication that humans use, which is storytelling.

So we are creating unique experiences where the machine generates an insight. And then, on the fly, we create data stories generated by the machine, thereby providing more context. As a user, you can have a great narrative, but then that narrative is expanded with insightful visualizations. From there, based on what you gain from the story, we are also looking at capabilities where you can explore further.

And in that third step you are still being augmented, but able to explore. It is user-driven. That is where you start introducing human intuition as well.

And when you think about the machine first surfacing insights, then getting more context with the data story, and lastly going to exploration — all three phases can be tied together in a seamless flow. You don’t lose the trust of the human. The context becomes really important. And you should be able to carry the context between all of the stages so that the user knows what the context is. Adding the human intuition expands that context.

Gardner: I really find this fascinating because we are talking not just about problem-solution, we are talking about problem-solution-resolution, then readjusting and examining the problem for even more solution and resolution. We are also now, of course, in the era of augmented reality, where we can bring these types of data analysis outputs to people on a factory floor, wearing different types of visual and audio cue devices.

So the combination of augmented reality, augmented intelligence, storytelling, and bringing it out to the field strikes me as something really unprecedented. Is that the case? Are we charting an entirely new course here?

Tutuk: Yes, I think so. It’s an exciting time for us. I am glad that you pointed out the augmented reality because it’s another research area that we are looking at. One of the research projects we have done augments people on retail store floors, the employees.

The idea is, if you are trying to do shelf arrangement, for example, we can provide them information — right when they look at the product – about that product and what other products are being sold together. Then, right away at that moment, they are being augmented and they will make a decision. It’s an extremely exciting time for us, yes.

Gardner: It throws the idea of batch-processing out the window. You used to have to run the data, come up with report, and then adjust your inventory. This gets directly to the interaction with the end-consumer in mind and allows for entirely new types of insights and value.

Tutuk: As part of that project, we also allow for being able to pin things on the space. So imagine that you are in a warehouse, looking at a product, an
d you develop an interesting insight. Now you can just pin it on the space on that product. And as you do that on different products, you can take a step back, take a look, and discover different insights on the product.

The idea is having a tray that you carry with you, like your own analytics coming with you, and when you find something interesting that matches with the tray – with, for example, the product that you are looking at — you can pin it. It’s like having a virtual board with products and with the analytics being augmented reality.

Gardner: We shouldn’t lose track that we are often talking about billions of rows of data supporting this type of activity, and that new data sets can be brought to bear on a problem very rapidly.

Putting data in context with AI

Tutuk: Exactly, and this is where our Associative Big Data Index technology comes into play. We are bringing the power of our unique associative engine to massive datasets. And, of course, with the latest acquisition that we have done with Attunity, we gain data streaming and real-time analytics.

Gardner: Digging down to the architecture to better understand how it works, the Qlik cognitive engine increasingly works with context awareness. I have heard this referred to as AI squared. What do you all mean by AI squared?

Tutuk: AI squared is augmented intelligence powered by an associative index. So augmented intelligence is our vision for the use of artificial intelligence, where the goal is to augment the human, not to replace them. And now we are making sure that we have the unique component in terms of our associative index as well.

Allow me to explain the advantage of the associative index. One of the challenges for using AI and machine learning is bias. The system has bias because it doesn’t have access to all of the data.

With the associative index, our technology provides a system with visibility to all of the data at any point, including the data that is associated with your context, and also what’s not associated. That part provides a good learning source for the algorithms that we are using.

For example, you maybe are trying to make a prediction for churn analysis in the western sales region. Normally if you select the west region the system — if the AI is running with a SQL or relational database — it will only have access to that slice of data. It will never have the chance to learn what is not associated, such as the customers from the other regions, to look at their behavior.

With the associative index, our technology provides a system with visibility to all of the data at any point, including the data that is associated with your context, and also what’s not associated. And that part that is not associated provides a good learning source for the algorithms that we are using. This is where we are differentiating ourselves and providing unique insights to our users that will be very hard to get with an AI tool that works only with SQL and relational data structures.

Gardner: Not only is Qlik is working on such next-generation architectures, you are also undertaking a larger learning process with the Data Literacy Program to, in a sense, make the audience more receptive to the technology and its power.

Please explain, as we move through this process of making intelligence accessible and actionable, how we can also make democratization of analytics possible through education and culturally rethinking the process.

Data literacy drives cognitive engine

Tutuk: Data literacy is important to help make people able to read, analyze, and argue with the data. We have an open program — so you don’t have to be a Qlik customer. It’s now available. Our goal is to make everyone data literate. And through that program you can firstly understand the data literacy level of your organization. We have some free tests you can take, and then based on that need we have materials to help people to become data literate.

As we build the technology, our vision with AI is to make the analytics platform much easier to use in a trusted way. So that’s why our vision is not only focused on prescriptive probabilities, it’s focused on the whole analytics workflow — from data acquisition, to visualization, exploration, and sharing. You should always be augmented by the system.

We are at just the beginning of our cognitive framework journey. We introduced Qlik Cognitive Engine last year, and since then we have exposed more features from the framework in different parts of the product, such as on the data preparation. Our users, for example, get suggestions on the best way of associating data coming from different data sources.

And, of course, on the visualization part and dashboarding, we have visual insights, where the Cognitive Engine right away suggests insights. And now we are adding natural language capabilities on top of that, so you can literally conversationally interact with the data. More things will be coming on that.

Gardner: As an interviewer, as you can imagine, I am very fond of the Socratic process of questioning and then reexamining. It strikes me that what you are doing with storytelling is similar to a Socratic learning process. You had an acquisition recently that led to the Qlik Insight Bot, which to me is like interviewing your data analysis universe, and then being able to continue to query, and generate newer types of responses.

Tell us about how the Qlik Insight Bot works and why that back-and-forth interaction process is so powerful.

Tutuk: We believe any experiences you have with the system should be in the form of a conversation, it should have a conversational nature. There’s a unique thing about human-to-human conversation – just as we are having this conversation. I know that we are talking about AI and analytics. You don’t have to tell me that as we are talking. We know we are having a conversation about that.

That is exactly what we have achieved with the Qlik Insight Bot technology. As you ask questions to the Qlik Insight Bot, it is keeping track of the context. You don’t have to reiterate the context and ask the question with the context. And that is also a unique differentiator when you compare that experience to just having a search box, because when you use Google, it doesn’t, for example, keep the context. So that’s one of the important things for us to be able to keep — to have a conversation that allows the system to keep the context.

Gardner: Moving to the practical world of businesses today, we see a lot of use of Slack and Microsoft Teams. As people are using these to collaborate and organize work, it seems to me that presents an opportunity to bring in some of this human-level cognitive interaction and conversational storytelling.

Do you have any examples of organizations implementing this with things like Slack and Teams?

Collaborate to improve processes

Tutuk: You are on the right track. The goal is to provide insights wherever and however you work. And, as you know, there is a big trend in terms of collaboration. People are using Slack instead of just emailing, right?

So, the Qlik Insight Bot is available with an integration to Microsoft Teams, Slack, and Skype. We know this is where the conversations are happening. If you are having a conversation with a colleague on Slack and neither of the parties know the answer, then right away they can just continue their conversation by including Qlik Insight Bot and be powered with the Cognitive Engine insights that they can make decisions with right away.

Gardner: Before we close out, let’s look to the future. Where do you take this next, particularly in regard to process? We also hear a lot these days about robotic process automation (RPA). There is a lot of AI being applied to how processes can be improved and allowing people to do what they do best.

The Qlik insight Bot is available with an integration to Microsoft Teams, Slack, and Skype. We know this is where the conversations are happening. They can just continue their conversation by including the Qlik Insight Bot and be powered with the Cognitive Engine insights that they can make decisions with.

Do you see an opportunity for the RPA side of AI and what you are all doing with augmented intelligence and the human cognitive interactions somehow reinforcing one another?

Tutuk: We realized with RPA processes that there are challenges with the data there as well. It’s not only about the human and the interaction of the human with the automation. Every process automation generates data. And one of the things that I believe is missing right now is to have a full view on the full automation process. You may have 65 different robots automating different parts of a process, but how do you provide the human a 360-degree view of how the process is performing overall?

A platform can gather associated data from different robots and then provide the human a 360-degree view of what’s going on in the processes. Then that human can make decisions, again, because as humans we are very good at making decisions by seeing nonlinear connections. Feeding the right data to us to be able to use that capability is very important, and our platform provides that.

Gardner: Elif, for organizations looking to take advantage of all of this, what should they be doing now to get ready? To set the foundation, build the right environment, what should enterprises be doing to be in the best position to leverage and exploit these capabilities in the coming years?

Replace repetitive processes

Tutuk: Look for the processes that are repetitive. Those aren’t the right places to use unique human capabilities. Determine those repetitive processes and start to replace them with machines and automation.

Then make sure that whatever data that they are feeding into this is trustable and comes from a governed environment. The data generated by those processes should be governed as well. So have a governance mechanism around those processes.

I also believe there will be new opportunities for new jobs and new ideas that the humans will be able to start doing. We are at an exciting new era. It’s a good time to find the right places to use human intelligence and creativity just as more automation will happen for repetitive tasks. It’s an incredible and exciting time. It will be great.

Gardner: These strike me as some of the most powerful tools ever created in human history, up there with first wheel and other things that transformed our existence and our quality of life. It is very exciting.

<br />

Listen to the podcast. Find it on iTunes. Read a full transcript or download a copy. Sponsor: Qlik.