FRAMEWORK : Data Analytics

Why Algorithms Will Never Be Better than Humans

August 3, 2018

In 2014, Stephen Hawking proclaimed AI could threaten our future: “Humans, limited by slow biological evolution, couldn’t compete and would be superseded by A.I.” [sic] That statement has been seconded by Bill Gates and Elon Musk.

This article takes a different view. Altman believes the reason algorithms can’t replace people is summed up in one word: maven. It comes from the Hebrew mēvīn for “person with understanding, teacher,” and is a participle of the verb hēvīn for “understanding.” To me, this is understanding at a deep level.

Altman believes that mavens long to share their knowledge. This differs from algorithms, which are a set of rules written to solve problems. Then he points out several other differentiating characteristics.

Mavens are curious. They ask questions and listen to understand. They also have personality: something that algorithms have a hard time scaling.

That’s because algorithms follow a different path. They achieve understanding through inputs. No matter how innovative an algorithm is, it can only produce the results that it determines. This can be limited by AI’s inability to pick up on the multitude of body language and voice modulation cues that a maven can – and get a sense when a person isn’t asking the question she really wants the answer to.

TripAdvisor and Netflix provide valuable information on vacation destinations or entertainment choices. We often supplement our online research by asking a friend — our personal maven — who has experienced what we’re exploring. Ultimately, we use both sets of information from our trusted sources — machine and human!

The relationship between AI and humans is complementary. AI presents us with a breadth of information analyzed and correlated. People provide “depth” — a deep dive tailored to our interests and delivered with passion. This combination of breadth and depth helps us discover and create solutions which are truly differentiated.

FRAMEWORK : Data Analytics

Smarter together: Why artificial intelligence needs human-centered design

August 3, 2018

Tay was a chatbot that learned through conversations with users. Twenty-four hours after it went online, the designers shut it down. That’s because pranksters trained it to utter racist, biased and fascist statements.

This raises an important difference between AI and people. We have common sense, which is proving elusive to teach machines. Tay wasn’t a maven and couldn’t understand the skewed input it was receiving. (For more on the link between common sense and emotional intelligence, see Think About the Human Touch in Simplifying Decisions, Using Big Data and Teaching Machines to Learn.)

AI can more efficiently and effectively perform certain tasks than people, such as reviewing a mass of customer complaint data and identifying emerging trends. AI contributes scale, speed and accuracy — which is important in the success of any business.

Algorithms — courtesy of big data and machine learning — are programmed to capture and transmit meaning. This is very different from the human qualities of being able to understand and originate meaning.

“Technology is the easy part. The hard part is figuring out the social and institutional structures around the technology.”  – John Seely Brown

Human-centered design creates a system that enables the best from both AI and people. Truly productive collaborations meet these four requirements:

  • Goal Relevance: What a computer sees as an optimal design must be tempered by the reality of how people use the system.
  • Handoff: People need to intervene during situations that are ambiguous, require common sense, or need context.
  • Feedback Loops: When creating algorithms that are fast and accurate, we must watch for any user and societal bias programmed into the data.
  • Psychological Impact: Just as we look at our effect on AI, we need to monitor its effect on us.

How can you make human-centered design relevant to you? Give your people and teams the leeway to share how they explore and understand their data, analysis and results. Ask for their straightforward assessment of what makes sense and what may require judgmental overlays — and why. Make them feel like mavens — so your business reaps the benefits.

“To grow your business, you must recognize your biggest challenge as a leader. This is to reward curiosity and common sense when embracing AI and other future-forward technologies.” – Marcia Tal

FRAMEWORK : Data Analytics

Top 5 Predictions for Data Quality in 2018

January 12, 2018

In “Top 5 Predictions for Data Quality in 2018,” Farnaz Erfan provides a financial frame for data quality trends: “Recent Gartner research indicates that the average financial impact of poor data quality (DQ) on organizations is $9.7 million per year.”

Now, we have a broader and deeper understanding that data—as raw material—needs to be governed and its quality measured. Importantly, this responsibility is broadening beyond centralized IT organizations to the lines of business. These businesses best understand the importance of their data quality as a driver to the business results for which they are accountable.

Get a glimpse of how IT, business domain experts, and data analysts will interact and collaborate in the future.

FRAMEWORK : Data Analytics

10 Predictions For AI, Big Data, And Analytics in 2018

January 12, 2018

Based on the Forrester Research report, “Predictions 2018: The Honeymoon Is Over,” Gil Press’ Forbes article “10 Predictions For AI, Big Data, And Analytics in 2018“ presents a statistical look into evolving trends.

Among the 10, a few highlights:

We see continued enhancements, interfaces, processes and discipline in integrating AI, NLP and conversational technologies, and real-time data visualizations. Further refined human and machine collaborations will continue to grow.

Analysts and business leaders will experience point and click analytics with conversational interfaces and artificial intelligence engines directing them in real time to guide better decision making.

See all 10 predictions.

FRAMEWORK : Data Analytics

2018 Top 10 Business Intelligence Trends with Videos

January 12, 2018

In “2018 Top 10 Business Intelligence Trends,” Tableau presents a series of ten well-illustrated posts—each with a short video from an expert.

One of the most interesting, “Liberal Arts Impact,” hits on one of our common themes in the last year:

“As analytics evolves to be more art and less science, the focus has shifted from simply delivering the data to crafting data-driven stories that inevitably lead to decisions. Organizations are embracing data at a much larger scale than ever before and the natural progression means more of an emphasis on storytelling and shaping data. The golden age of data storytelling is upon us …”

“The NLP Promise” ranks as another top trend:

Consumers, analysts, developers and engineers will be benefitting from Natural Language Processing (NLP). Using voice data and studying the ways in which we ask questions will enhance our ability to expand our understanding of consumers and improve business outcomes. 

Watch all 10 short videos.

FRAMEWORK : Data Analytics

Top 10 Trends for Digital Transformation In 2018

January 12, 2018

Change itself is the major trend in 2018—as in any age of transformation. In the ongoing Digital Transformation, Daniel Newman posts “Top 10 Trends for Digital Transformation in 2018.”

Digital transformation will continue through the IoT:

“The mass amount of information being created by the IoT has the power to revolutionize everything from manufacturing and healthcare to the layout and functioning of entire cities — allowing them to work more efficiently and profitably than ever before.”

Specifically, tech giants are seeing the power of combining IoT and Analytics—IoT Analytics— “driving new business insights across a vast array of industries.”

See more trends, including AI, Blockchain, Edge Computing and 5G—and an infographic showing how each contributes to the ongoing Digital Transformation.

FRAMEWORK : Data Analytics

Data & Analytics Trends in 2018: What Do the Experts Expect?

January 12, 2018

Ben Davis and Econsultacy give us a look at what the experts predict for data and analytics in 2018.

In addition to optimizing marketing campaigns, enhancing the customer experience using artificial intelligence and machine learning will continue to be core to brands’ personalized messaging and human experiences. The use of integrated data—including attribution and campaign results with analytic capabilities and a relentless focus on the customer experience—will be key.

Compliance will continue to provide challenges and opportunities as data models, analytic roadmaps and strategies are reviewed and governed. Taking effect in May 2018, the EU General Data Protection Regulation (GDPR) will have a significant impact on tracking and targeting.

See more expert predictions for prescriptive analytics, data visualization, data integration, and “data for the masses.”

FRAMEWORK : Data Analytics

5 Artificial Intelligence Predictions For 2018

January 12, 2018

In 5 Artificial Intelligence Predictions For 2018, Daniel Newman moves us away from “bigtime futuristic predictions of AI” and provides a more realistic view of how artificial Intelligence will continue to be an even greater part of our personal and professional lives.

Data-driven machines will alter work streams between humans and machines to drive efficiencies and effectiveness. Conversational technologies, “listening to conversations,” and smart automation will continue to improve and enhance both our personal and professional lives.

Read more AI predictions here.

FRAMEWORK : Data Analytics

Will Blockchain Transform Financial Services?

December 29, 2017

Can blockchain technology transform the financial services industry?

Authors Joichi Ito, Neha Narula and Robleh Ali explore this question by drawing a parallel with the Internet in their article “The Blockchain Will Do to the Financial System What the Internet Did to Media.”

In the 1990s, many believed the Internet to be a fad. In fact, “the mainstream press scoffed when Nicholas Negroponte predicted that most of us would soon be reading our news online rather than in a newspaper.”

We all know how prescient Negroponte’s prediction was. The authors say email was the “killer app for the early internet;” continuing with their comparison, they believe that Bitcoin is “the killer app for the blockchain.”

However, for Jamie Dimon “bitcoin the currency, I think is going nowhere…the blockchain is a technology which we’ve been studying and, yes, it’s real.” (CNBC interview). He foresees broad commercial and banking applications “if it proves to be cheap and secure.”

As all things in the financial services industry, we should recognize the importance of commercial applications for blockchain technology at the beginning—as opposed to the unrestrained development of the Internet.

When we look forward 20 years what will we see?

Blockchain is truly an innovative—even transformational—technology.

The important thing about blockchain is that it forces you to write rules. That’s what banks do best: write business rules, write policies, rewrite rules, rewrite policies—all on very complex transaction-based systems.

Think about news agencies, music companies, advertising, television, and movies: all businesses forever changed by the Internet. If we can prove its value, fast-forward 20 years and blockchain will be mainstream in the financial services industry—whatever mainstream will look like then.

FRAMEWORK : Data Analytics

What's So Smart About Artificial Intelligence?

December 29, 2017

Try asking Siri: “What does “AI” mean?”

Ian Bogost more successfully tackles that question in his Atlantic article, “‘Artificial Intelligence’ Has Become Meaningless.” Distinguishing what he calls “supposed-AI” from real artificial intelligence, Bogost searches for a meaningful definition.

The author, of course, doesn’t suggest that artificial intelligence itself is meaningless but that the term “AI” escapes clear meaning by imprecise usage and exponential overuse.

From Google to Facebook…from the Homeland Security to Coca-Cola…from earnings call transcripts to corporate strategy to press releases—examples of “supposed AI” abound. Bogost believes that most systems “making claims to artificial intelligence aren’t sentient, self-aware, volitional, or even surprising. They’re just software.”

To gain clarity, Georgia Tech artificial intelligence researcher Charles Isbell defines AI this way: “Making computers act like they do in the movies.”

Not to seem “too glib,” Isbell then identifies two features “necessary before a system deserves the name AI.”

First, the system “must learn over time in response to changes in its environment.”

Second, “what it learns to do must be interesting enough that it takes humans some effort to learn.”

Real AI is not just repeatable standard process; rather, it’s the learning of things as we go through the process of learning itself. This method includes interface, integration, conversation—all the things that human beings experience in learning and applying what they learn. It’s this learning that differentiates artificial intelligence from “mere computational automation.”

The same kind of imprecise terminology and clarity of definition affects our understanding of “analytics.” Often used as an umbrella term, “analytics” encompasses many skill sets, techniques, specializations, and tools along its continuum.

Think of the commonly defined analytical processes:

  • Descriptive – what happened
  • Diagnostic – why something happened
  • Discovery – learning and insight
  • Predictive – likely to happen
  • Prescriptive – recommended action

We use all relevant techniques, tools and technologies available to perform these analytical processes. In turn, the application for these analytic processes vary depending on the business need. Some examples of business needs which require different expertise and tools are:

  • Information solutions—data queries to advanced reporting
  • Business solutions—campaign tracking to predictive response modeling
  • Quantitative solutions—decision trees to game theory

Running the gamut from reporting to cognitive computing, analytics needs to be viewed and discussed in ways that appreciate the complexity, applications, skill sets, value, and the human element.

What do AI and analytics have in common? Look to the core of analytics to find a meaningful definition for “AI”—“machinery that learns and then acts on that learning.” Just like in the movies.

Photo Credit: Gordon Tarpley from PS-Hoth Set

FRAMEWORK : Data Analytics

How Companies Can Use Customer Data to Create Value

December 15, 2017

Is it self-serving to recommend an article because it reflects the essence of our business? No, I believe so strongly in creating value from your own data that I’ll take that risk.

In the McKinsey&Company article, Capturing value from your customer data,” Brad Brown, et al. make a strong case for companies to “put their information to work” in creating financial impact:

“The ability to capture and use customer insights to shape products, solutions, and the buying experience as a whole, is critically important. Research tells us that organizations that leverage customer behavioral insights outperform peers by 85 percent in sales growth and more than 25 percent in gross margin.”

Further, a McKinsey survey showed that spending on analytics “to gain competitive value, to target customers more successfully, and to optimize operations and supply chains generated operating-profit increases in the 6 percent range.”

In our experience, creating significant financial impact starts with a company’s own data. That’s the low-hanging fruit, and there’s a lot of it to be had—the value companies can discover in data that is just waiting to be extracted.

How does a company use customer data to create value?

Define desired outcomes. It’s important to also define exactly how those outcomes can be realized.

Enrich customer data. Use outside resources to enrich data and make that data accessible and sharable across the enterprise.

Use internal assets. Use client relationships, deep business expertise and scale to build capabilities.

Identify patterns, behaviors and insights. And then apply them to solving problems and creating new solutions.

Improve productivity. Use data-enabled processes and optimization to drive productivity gains.

What’s the biggest hurdle to capturing value?

The authors emphasize that “breakthroughs don’t have to be the preserve of digital pure plays. Many incumbent organizations have the advantage of long-standing client relationships, deep pockets of expertise, and scale.”

We all agree, however, that you can’t skip defining specific outcomes. The authors recommend “prioritizing a handful of specific customer outcomes, such as reduced churn or improved cross-sell, and setting up small, dedicated cross-functional teams to experiment, refine and release new approaches.”

Without defined outcomes—and knowing how your people, processes, and systems work together to implement testing and rollouts—your data analysis and insight will remain only academic.

The real power comes from joining learning and financial impact to create value for companies and customers alike. 

FRAMEWORK : Data Analytics

Why Fintech and Banks Complement Each Other

October 27, 2017

Can’t you just hear Gilda Radner’s character Emily Litella saying, “What’s all this fuss I hear about financial eruption?”

According to Chris Skinner’s American Banker article “The battle between banks and disruptors is just beginning,” Fintechs believe they can exploit big banks’ legacy weaknesses and draw customers to the digital banking world.

Skinner cites Stripe as an example, as the six-year old startup is the “preferred code for building online checkout services.” Valued at over $10 billion in 2016, Stripe has “has taken something that banks make difficult—setting up online payment services—and made it incredibly easy for customers.”

You only have to frequent a farmers’ market to see almost all the produce and craft vendors using Stripe or Square for credit or debit card payment via smartphones. Digital services provide small business an easy, secure and remote-friendly way to accept bank issued credit or debit cards for payment.

Is this disruption or is it complementary services working together to serve the customer in new ways?

As Skinner comments, the most “successful fintech firms are not replacing banks…they are serving markets that were underserved…succeeding by addressing areas that banks find difficult to serve due to cost or risk.”

The opportunity for everyone is for banks and fintech firms to collaborate on services that banks don’t or won’t do well, or where banks are held back by legacy data systems, institutional process, or regulation. Fintech’s unbiased thinking, digital prowess and cost efficiency will bring new reach and service to these underserved areas, making it a win-win for all involved, most importantly, customers.

The future of banking is not so much a “fight between a host of new digital players and a few large banks that find it hard to change.” The future of banking will be a collaborative adaption among established banks and emerging digital fintech companies.

So, what’s all this we hear about financial eruption?

Never mind.