Sections

Finding The Humanity In Data: IBM Watson’s CTO Rob High Defines The Emerging Era Of Cognitive Computing [PODCAST]

Hessie Jones

It’s increasingly clear that we, as humans, continuously upload our identities every day. Those needs and propensities become quantified and contextualized. For companies like IBM Watson, understanding the human condition is important so technology can increasingly define patterns, learn, and potentially predict outcomes that benefit both business and industry.

We were pleased to host Rob High, IBM Fellow, vice president and chief technology officer, Watson Solutions, IBM Software Group. In this episode, Rob talks about:

  • The definition of cognitive computing
  • How Watson is aiding the advancement of health care
  • Chef Watson and recommendation on recipes
  • Advanced cognitive systems and how they’re applied across different mediums
  • The future of AI – Should humans be fearful?

You can listen to the podcast here or catch the episode here on Libsyn.

What is cognitive computing?

  • Ultimately, cognitive computing has the greatest benefit for people. By definition, it is the interpretation of the human condition that includes all those things that we take in every day: the information, our communication. It deciphers the intent we derive from them that is meaningful and used in the way we make decisions in our everyday lives.
  • Cognitive computing augments our own human cognition and gives us the insight and inspiration to those specific things we need to know to do our job better.

Classical computing methods have been unable to understand the underlying intent in how we, as humans, have communicated with each other, through voice and text, audible or written.

Cognitive computing does not replace human thinking. It does the research for you so you can do your thinking better.

How is Watson making strides in healthcare?

Watson can operate only in digital form, aggregating the information and looking at discreet elements to explicitly understand the various treatment options to better inform decision making.  Massive amounts of data will uncover trends across the population and yield certain correlations that may help interpret and predict patient response to various treatments.

Through Watson’s work with MD Anderson Cancer Centre, the Oncology Expert Advisor (OEA) was launched.

By pulling together and analyzing vast amounts of information from patient and research databases, the OEA is expected to help our care teams identify and fine-tune the best possible cancer treatments for our patients, while also alerting them to problems that arise during a patient’s care.

In accessing millions of patient records, Watson can aid in identifying a micro-segmentation of the population that have common traits; i.e., exposure to environmental impacts, genetics, heritage, and symptoms. These will aid in surfacing the opportunities to apply the knowledge and understanding to determine how well someone with the same exposure will respond to certain treatments.

While health information across the world has been fragmented, Watson can aid in processing massive quantities of information (not humanly possible) to create implications in a meaningful way, and in a short period of time. Now doctors and patient caregivers who have documented success can share that information with other medical practitioners across the globe to accelerate diagnosis and treatment.

Chef Watson: “Ready to do some cognitive cooking?”

This was for me the most fascinating part of the segment: Chef Watson enables people to make decisions about menus, identifying and helping us discover new recipes based on our unique preferences.

At IBMchefwatson.com,Watson partnered with Bon Appetit, which provided 9000 recipes for Watson to ingest and learn about the different types and styles of recipes. For a computer which innately has no sense of palette or smell, Watson learned about the taste makeup and flavors and the feeling that results when you consume a particular dish. It also learns about the science of taste chemistry and the chemical compounds that give the recipes their specific tastes. From this perspective, it has the ability to begin to imitate the human senses. As per Rob:

Watson starts from scratch, dealing with many – potentially up to a quintillion– combinations of ingredients when it comes up with its unique recommendation every time.

It’s getting at the root of what makes people who they are – the things we experience are interpretable.

As an example, if you wanted a Belgian flavor for a given recipe, Watson will evaluate the different combinations of ingredients that pair well and produce a Belgian flavor, and may come up with different variations.

Starting out as a fun and interesting project, this has occurred as a result of the cognitive ability and has allowed Watson to venture into the art of the possible.

Patterns and the evolution of interpretations

Similar to the learnings with MD Anderson, there are trends or patterns within the data where we can derive the greatest understanding or intention. Overlay contextual history which informs more of the human understanding. Collectively these allow us to extract meaning. Cognitive systems draw meaning that can bring the right set of information to humans and attention to just the right thing(s) to shape the decision-making process.

Pervasive technology has been able to to process 20% of the world’s information until now. The other 80% of that data is the human condition: the spoken word, written word, music, visual representations – all interpretations of our interests and needs. This is the heart of understanding. As Rob points out:

Multi-modal is how we communicate with each other: Not only what you’re hearing, but the intonation in the voice reflects the substance of that expression that’s being conveyed. Add the cadence that punctuates these points and now we know how humans understand each other. The computer needs to understand that as well.

Cognitive systems are not based on the same mathematical models as traditional computers. Attempting to interpret the human condition is doing so in the presence of idiosyncrasies and nuances carried through conversations and other communications.

Our words, our expressions are ambiguous…

Are these models reliable?

There is “no absolute level of correctness necessary;” these results are being applied in the eyes of the beholder. The computer will need to be exposed to enough examples that it will begin to surface patterns of meaning that will allow it to work well in that context. Be prepared for the outcomes to vary by environment or time period or when new variables are introduced.

What is the future of AI? Should we, as humans, be fearful?

The potential of cognitive is vast and in the near future, the amazing strides that are introduced are evidence of the inherent benefit to our human strength and potential.

Technology will continue to progress and there will always be a risk that people and organizations will use it in nefarious ways.

Technology should not be feared. With increased understanding comes progress. It also means humans should be responsible and use it for the purposes for which it was intended.

As this information becomes for common, technology companies need to ensure safeguards are put in place to mitigate abuse to our privacy.

Rob High is an IBM Fellow, vice president and chief technology officer, Watson Solutions, IBM Software Group. He has overall responsibility to drive Watson Solutions technical strategy and thought leadership. As a key member of the Watson Solutions Leadership team, Rob works collaboratively with the Watson engineering, research, and development teams across IBM.

Want more on future tech and its effect on business? See Bring Your Robot To Work.

Comments

Smart Machines Create Markets For Cyber-Physical Advances

Marion Heindenreich

Today, industrial machines are more intelligent than ever before. These intelligent machines are changing companies in many ways.

Why smart machines?

Mobile networked computers were a key breakthrough for making smart machines. Big Data allows machines and computers to store information and analyze complex patterns. Cloud computing offers broad access to information and more storage.

These computerized machines are both physical and virtual. Some call them “cyber-physical” machines. Technology lets them be self-aware and connected to each other and larger systems.

Businesses change their approaches

Intelligent machines allow companies to innovate in many areas. For one, the value proposition for customers is evolving. Businesses now model and plan in different ways in many industries.

Makers of industrial machines and parts work in new ways within the organization. Engineering now partners with mechanical, electronic, and software staff to develop new products. Manufacturing now seamlessly ties what happens on the shop floor to the customer.

Service models are changing too. Scheduled and reactionary servicing of machines is fading. Now intelligent machines track themselves. Machines detect problems and report them automatically. Major problems or failures are predicted and reported.

A data mining example

One good industrial example is mining, which can be dangerous and difficult. As ores become scarce, the costs of mining have increased.

“Smart machines” started in mining in the late 1990s. Software and hardware let remote users change settings. Operators moved hydraulic levers from a safe distance. Sensors observed performance and diagnosed issues.

Data cables connected machines to computers on the surface. Continuous and remote monitoring of the machines grew. Over time, embedded sensors helped improve monitoring, diagnostics, and data storage.

The technology means workers only go underground to fix specific issues. As a result, accident and injury risk is lower.

New wireless technology now lets mining companies connect data from many mine sites. Service centers access large amounts of data and can improve performance. Maintenance is prioritized and equipment downtime is reduced.

Opportunity abounds

For companies the time is now. Today, mobile “connected things” generate 17% of the digital universe. By 2020 that share grows to 27%.

You might not be investing in this so-called “Internet of Things” (devices that connect to each other). But it’s a good bet your competitors are. A December 2015 study reported 33% of industrial companies are investing in the Internet of Things. Another 25% are considering it.

There are risks

This new dawning era of manufacturing is exciting. But there are concerns. Cyber attacks on the Internet of Things are not new. But as the use of intelligent machines grows, the threat of cyber attacks in industry grows.

Data confidentiality and privacy are concerns. So too are software and hardware vulnerabilities. Exposure to attack lies not just in the virtual space but the physical too. Tampering with unattended machines and theft pose serious risk.

To address these threats, industries must invest in cybersecurity along with smart machines.

Conclusion

The potential advantages of smart machines are staggering. They can reshape industries and change how companies produce new products and create new markets.

For more information, please download the white paper Digital Manufacturing: Powering the Fourth Industrial Revolution.

Comments

Marion Heindenreich

About Marion Heindenreich

Marion Heidenreich is a solution manager for the SAP Industrial Machinery and Components Business Unit who focuses on solution innovations like Product Costing on SAP HANA and cloud solutions, as well as providing financial and business analysis for industry business strategy definition and business planning.

Mining Firms Turn To Tech

Ruediger Schroedter

Gone are the days in mining when assessments of potential dig sites meant lots of waiting for results. Gone, too, is the uncertainty on a mine job about where to go next.

For mining executives, recent advances in digital technology allow companies to make decisions at a rapid pace. Decisions that used to take days and weeks now can be done in minutes and hours.

With more information available faster, mining leaders reduce both short- and long-term financial risk. Data from across the enterprise inform decisions about buying and selling assets. Profitability should increase, driven by key technology advances.

Digging in to the data

There are two key drivers to this digital revolution. The first is the rise of the Internet of Things (IoT). The IoT consists of devices that are equipped with sensors, software, and wireless capabilities. These devices are connected to each other and can detect, store, and send data.

Bonus: Click here to learn more about Digital Transformation in Mining.

The second is the rise of Big Data, mobile, and cloud computing. Today’s mobile devices can track, send, and receive data from remote sites worldwide. Cloud computing stores billions of bytes of data at low cost. Big Data analytics programs take data coming from many different locations and systems and synthesize it. Those programs then better inform decisions by offering dashboards, metrics, and predictive modeling.

Robots are able to venture into hazardous areas and move material with remote human oversight. On-site mining data is sent via mobile phone to a cloud-based platform. For mining, the convergence of these technologies provides extraordinary possibilities.

Technology at play

The potential impact is significant. A recent report by McKinsey & Co. showed the use of advanced analytics in mining and related industries had a major impact. Firms using these programs to assess production areas increased their profit margins by 2-3 percentage points.

One mining company used so-called Monte Carlo simulations to reduce certain capital expenses. Monte Carlo simulations use complex algorithms and repeated random sampling to model possible outcomes. They’re frequently used in finance, biology, and insurance. The Mining Journal reported how the company challenged assumptions about a project’s capital needs. It took historical data on certain disruptions such as rainfall patterns. Then models of its mines were made showing the impact of flooding and rainwater. The data led to a new strategy that maximized storage capacity and handling across all its mines. Capital costs dropped by 20 percent.

18 Aug 2012, South Dakota, USA --- USA, South Dakota, Lead, View of open pit --- Image by © Bryan Mullennix/Tetra Images/Corbis

Buy or sell?

With so many variables at play, mining valuation is not for the faint of heart. Integrated data streams available at the discovery stage make for better informed purchase decisions.

Software programs today can take data to build and validate exploration models. These programs use 3D visualization and validated geophysical, analytical, and drill hole data. In turn, detailed 3D topographical models are possible.

Other programs assess historical, assay, and drilling data. This information creates viable scenarios for determining whether to buy or sell a site.

These tools use data consistently from one potential site to the next, allowing for forecasting of economic risk that is consistent across the organization. The firm today can use “real options valuation” to develop models of outcomes given changing economic conditions. With clearer information about potential risks, firms can decide whether to stage, sell, abandon, expand, or buy.

Anticipating, not reacting

Mining companies realize today that these analytic platforms and dashboards offer many advantages. Users have a clearer interpretation of the aggregated and analyzed data points from multiple areas. Using predictive analytics, mining decisions are made based on smart assumptions, not past historical information.

Robust software programs can generate reports almost instantaneously. Supervisors have on-site access to the analysis through a web browser or app. This data has many uses. Drilling managers save time and can make quicker decisions on next moves. Supplies can be ordered faster. Needed data for accreditation and compliance is immediately accessible.

Selecting the right sites

One example is assay analysis. Today, geologists do not wait weeks or months for assay results. Instead of off-site analysis, web-based applications deliver information much faster to inform decisions.

Robots are sending information about field operations, safety, needed maintenance, and drilling performance.  Some devices send the information themselves. In other cases, staff use mobile phones, tablets, or laptops.  This information and analytics in turn help with site selection. Integrating data from mine planning, ventilation, safety, rock engineering, and mineral resources improves overall forecasting.

Discovery, particularly of Tier 1 sites, is an increasingly costly venture for mining companies. Demand for many products is increasing while discovery rates are dropping. Mined product is of a lesser quality, particularly in mature mining locations. Many possible sites are in areas that are underexplored areas with difficult and deep cover.

The advanced technologies available today are contributing to rapid improvement in these discovery issues.

Prospective drilling

Consider the drill hole. To reduce costs in exploration, there needs to be enough rich information from the opening drill hole. It needs to be delivered in as close to real time as possible. Doing so lessens the risk of the second drill hole. Better information from the start helps improve vectoring. It provides better information about what mineral systems are being drilled.

This approach, called prospective drilling, is becoming increasingly used in mining. It employs drilling activity to map covered mineral systems. In turn, geochemical and geophysical vectoring can lead firms toward deposits.

Australia has invested heavily in this area. The Deep Exploration Technologies Cooperative Research Centre (DET CRC) has a singular vision: uncovering the future. Its core purpose is “develop transformational technologies for successful mineral exploration through deep, barren cover rocks.”

To get to that point, the DET CRC is borrowing a drilling technique from the oil business. Coiled tubing is paired with downhole and top-of-the-hole sensors. The informaton provides petrophysical, structural, rock fabric, geochemical, and mineralogical data all at once.

Conclusion

To meet increasing demands for new viable sites, and to improve efficient on sites, mining is changing. Using smart, connected products and robust data modeling, mining is being done faster, safer, and more efficiently than ever.

Join a LiveTwitterChat on digitalization in mining on May 4th from 10-11 a.m. EST: #digitalmining

The global mining and metals industry will come together to discuss how digital innovation is impacting the mining industry July 12-14 at the International SAP Conference for Mining and Metals in Frankfurt, Germany.  Don’t miss this opportunity to meet with world leaders and learn how your organization can become a connected digital enterprise.

Follow speakers and pre-event activities by following sapmmconf and @sapmillmining on Twitter

AA Mining and Metals Forum

Comments

Ruediger Schroedter

About Ruediger Schroedter

Ruediger Schroedter is responsible for solution management of SAP solutions for the mining industry worldwide. He has spent more than 15 years in the mill products and mining industries and has extensive experience implementing SAP solutions for customers in these industries before coming to SAP.

How Much Will Digital Cannibalization Eat into Your Business?

Fawn Fitter

Former Cisco CEO John Chambers predicts that 40% of companies will crumble when they fail to complete a successful digital transformation.

These legacy companies may be trying to keep up with insurgent companies that are introducing disruptive technologies, but they’re being held back by the ease of doing business the way they always have – or by how vehemently their customers object to change.

Most organizations today know that they have to embrace innovation. The question is whether they can put a digital business model in place without damaging their existing business so badly that they don’t survive the transition. We gathered a panel of experts to discuss the fine line between disruption and destruction.

SAP_Disruption_QA_images2400x1600_3

qa_qIn 2011, when Netflix hiked prices and tried to split its streaming and DVD-bymail services, it lost 3.25% of its customer base and 75% of its market capitalization.²︐³ What can we learn from that?

Scott Anthony: That debacle shows that sometimes you can get ahead of your customers. The key is to manage things at the pace of the market, not at your internal speed. You need to know what your customers are looking for and what they’re willing to tolerate. Sometimes companies forget what their customers want and care about, and they try to push things on them before they’re ready.

R. “Ray” Wang: You need to be able to split your traditional business and your growth business so that you can focus on big shifts instead of moving the needle 2%. Netflix was responding to its customers – by deciding not to define its brand too narrowly.

qa_qDoes disruption always involve cannibalizing your own business?

Wang: You can’t design new experiences in existing systems. But you have to make sure you manage the revenue stream on the way down in the old business model while managing the growth of the new one.

Merijn Helle: Traditional brick-and-mortar stores are putting a lot of capital into digital initiatives that aren’t paying enough back yet in the form of online sales, and they’re cannibalizing their profits so they can deliver a single authentic experience. Customers don’t see channels, they see brands; and they want to interact with brands seamlessly in real time, regardless of channel or format.

Lars Bastian: In manufacturing, new technologies aren’t about disrupting your business model as much as they are about expanding it. Think about predictive maintenance, the ability to warn customers when the product they’ve purchased will need service. You’re not going to lose customers by introducing new processes. You have to add these digitized services to remain competitive.

qa_qIs cannibalizing your own business better or worse than losing market share to a more innovative competitor?

Michael Liebhold: You have to create that digital business and mandate it to grow. If you cannibalize the existing business, that’s just the price you have to pay.

Wang: Companies that cannibalize their own businesses are the ones that survive. If you don’t do it, someone else will. What we’re really talking about is “Why do you exist? Why does anyone want to buy from you?”

Anthony: I’m not sure that’s the right question. The fundamental question is what you’re using disruption to do. How do you use it to strengthen what you’re doing today, and what new things does it enable? I think you can get so consumed with all the changes that reconfigure what you’re doing today that you do only that. And if you do only that, your business becomes smaller, less significant, and less interesting.

qa_qSo how should companies think about smart disruption?

Anthony: Leaders have to reconfigure today and imagine tomorrow at the same time. It’s not either/or. Every disruptive threat has an equal, if not greater, opportunity. When disruption strikes, it’s a mistake only to feel the threat to your legacy business. It’s an opportunity to expand into a different marke.

SAP_Disruption_QA_images2400x1600_4Liebhold: It starts at the top. You can’t ask a CEO for an eight-figure budget to upgrade a cloud analytics system if the C-suite doesn’t understand the power of integrating data from across all the legacy systems. So the first task is to educate the senior team so it can approve the budgets.

Scott Underwood: Some of the most interesting questions are internal organizational questions, keeping people from feeling that their livelihoods are in danger or introducing ways to keep them engaged.

Leon Segal: Absolutely. If you want to enter a new market or introduce a new product, there’s a whole chain of stakeholders – including your own employees and the distribution chain. Their experiences are also new. Once you start looking for things that affect their experience, you can’t help doing it. You walk around the office and say, “That doesn’t look right, they don’t look happy. Maybe we should change that around.”

Fawn Fitter is a freelance writer specializing in business and technology. 

To learn more about how to disrupt your business without destroying it, read the in-depth report Digital Disruption: When to Cook the Golden Goose.

Download the PDF (1.2MB)

Comments

Tags:

Sherry Turkle: We Need to Talk

Stephanie Overby

reclaiming-conversation-sherry-turkle-200x300MIT psychologist Sherry Turkle on why we need to talk to our colleagues

Human beings are communicating more often and with more people than ever before, thanks to the digital devices we are all but tethered to. But the art of conversation is in decline. MIT psychologist Sherry Turkle, who has devoted her career to examining the impact of technology on human interaction, lays out some worrying consequences in her latest book, Reclaiming Conversation: The Power of Talk in a Digital Age. Overreliance on digital communication has not only affected our ability to have effective face-to-face exchanges but has also diminished our capacity for empathy and intimacy. In addition, digital discussions are often less productive and effective than in-person interactions.

We talked to Turkle about the value of human interaction that is unmediated by technology, when to choose talking over texts and e-mail, and how corporate leaders can revive conversation in the digital workplace.

Q: The big trend in business is digital transformation. A major goal is to automate and digitize interactions. What are companies losing in the bargain?

Sherry Turkle: When you want to build trust, when you want to get to know someone new, when you want to seal a deal—these are not moments for transactions, which are fairly blunt and objective instruments for communicating information. These are times for conversations, which are subjective and emotional and enable greater understanding. Good managers need to know when they are dealing with a moment when a transaction is appropriate and when it is a moment for a human exchange. If you try to be transactional when you need a conversation, you are on your way to frustration, disappointing results, and—most often—the need to do it all again.

Q: How has the increase in digital communications affected our ability to talk to each other?

Turkle: We find ways to not have the conversations that count. We would rather keep communication on screens. As one young man told me when I asked what was wrong with conversation: “It takes place in real time, and you can’t control what you’re going to say!” Of course, that is what’s “wrong” with conversation. But, it is also what’s profoundly right with conversation. It is a place where intimacy is born. The link between face-to-face conversation and empathy is strong. There has been a 40% decline in empathy among college students over the past 20 years, with most of that decline happening in the past decade.

Q: Why is face-to-face conversation important in business? Can’t that  effectively be simulated using technology?

Turkle: We are creatures designed for broadband, rich, nuanced exchange through our voices and faces. We are inventing new languages on the screen, and we are doing that with invention, wit, and nuance. But in business (as in friendship and love), we are misunderstanding each other—badly. And we are sending 10 e-mails where a brief call would do.

I am a pragmatist. When you need a video link or a call, use these tools. But what I see is people avoiding presence when it is possible.

Q: How can managers make a business case for talking?

Turkle: Research shows that conversation is good for the bottom line. People are more productive, creative, and engaged with their work when they have time for face. to-face talk. Sociologist Ben Waber had employees wear “sociometric badges” that measured their conversational patterns. When people were given coffee breaks together, performance improved. One CEO I interviewed instituted a breakfast meeting for his team. It gave them all an opportunity to share ideas and talk freely. Group productivity increased, and they needed fewer formal meetings.

One “easy” change is to eliminate devices from in-person meetings. The research is clear: devices distract. They diminish conversations and the relationships among participants. Make meetings shorter if necessary. Offer breaks. Designate one employee to notify attendees if an emergency arises. A meeting is a time to meet.

Q: What else can leaders do to encourage conversation amid the pressure to digitize?

Turkle: Make it clear that in your organization being online is not how you show your loyalty. Instead, show that what is valued is an employee who picks up the phone. Visit your colleagues in person. If you talk, others will talk. Also, design the workplace for conversation by creating device-free spaces that encourage it. Help employees work through their terror of real- time conversations by making it clear that revealing your thought process is valued. Finally, be less transactional. Begin an answer to an e-mail by saying, “I’m thinking.” It’s a powerful message. Complicated problems require thinking and then time to talk.

Q: We conducted this interview electronically to accommodate our schedules. What did I miss out on? How about you?

Turkle: We missed out on the chance to know each other better. What we had was a transaction. I took the time to lay out some of my ideas. But you and I are not closer for it. In business, this would not put us in the best relationship to move forward with a project. Now would be time for conversation!

 

Comments