Tag: Artificial Intelligence

Eighteen best practices for human-centered AI design

By Mihaela Vorvoreanu, Saleema Amershi, and Penny Collisson

Today we’re excited to share a set of Guidelines for Human-AI Interaction. These 18 guidelines can help you design AI systems and features that are more human-centered. Based on more than two decades of thinking and research, they have been validated through a rigorous study published in CHI 2019.

Why do we need guidelines for human-AI interaction?

While classic interaction guidelines hold with AI systems, attributes of AI services, including their accuracy, failure modalities, and understandability raise new challenges and opportunities. Consistency, for example, is a classic design guideline that advocates for predictable behaviors and minimizing unexpected changes. AI components, however, can be inconsistent because they may learn and adapt over time.

We need updated guidance on designing interactions with AI services that provide meaningful experiences, keeping the user in control and respecting users’ values, goals, and attention.

Why these guidelines?

AI-focused design guidance is blooming across UX conferences, the tech press, and within individual design teams. That’s exciting, but it can be hard to know where to start. We wanted to help with that, so…

• We didn’t just make these up! They come from more than 20 years of work. We read numerous research papers, magazine articles, and blog posts. We synthesized a great deal of knowledge acquired across the design community into a set of guidelines that apply to a wide range of AI products, are specific, and are observable at the UI level.
• We validated the guidelines through rigorous research. We tested the guidelines through three rounds of validation with UX and HCI experts. Based on their feedback, we iterated the guidelines until experts confirmed that they were clear and specific.

Let’s dive into the guidelines!

The guidelines are grouped into four categories that indicate when during a user’s interactions they apply: upon initial engagement with the system, during interaction, when the AI service guesses wrong, and over time.

Initially

1. Make clear what the system can do.

2. Make clear how well the system can do what it can do.

The guidelines in the first group are about setting expectations: What are the AI’s capabilities? What level of quality or error can a user expect? Over-promising can hurt perceptions of the AI service.

PowerPoint’s QuickStarter illustrates Guideline 1, Make clear what the system can do. QuickStarter is a feature that helps you build an outline. Notice how QuickStarter provides explanatory text and suggested topics that help you understand the feature’s capabilities.

During Interaction

3. Time services based on context.

4. Show contextually relevant information.

5. Match relevant social norms.

6. Mitigate social biases.

This subset of guidelines is about context. Whether it’s the larger social and cultural context or the local context of a user’s setting, current task, and attention, AI systems should take context into consideration.

AI systems make inferences about people and their needs, and those depend on context. When AI systems take proactive action, it’s important for them to behave in socially acceptable ways. To apply Guidelines 5 and 6 effectively, ensure your team has enough diversity to cover each other’s blind spots.

Acronyms in Word highlights Guideline 4, Show contextually relevant information. It displays the meaning of abbreviations employed in your own work environment relative to the current open document.

When Wrong

8. Support efficient dismissal.

9. Support efficient correction.

10. Scope services when in doubt.

11. Make clear why the system did what it did.

Most AI services have some rate of failure. The guidelines in this group recommend how an AI system should behave when it is wrong or uncertain, which will inevitably happen.

The system might not trigger when expected, or might trigger at the wrong time, so it should be easy to invoke (Guideline 7) and dismiss (Guideline 8). When the system is wrong, it should be easy to correct it (Guideline 9), and when it is uncertain, Guideline 10 suggests building in techniques for helping the user complete the task on their own. For example, the AI system can gracefully fade out, or ask the user for clarification.

Auto Alt Text automatically generates alt text for photographs by using intelligent services in the cloud. It illustrates Guideline 9, Support efficient correction, because automatic descriptions can be easily modified by clicking the Alt Text button in the ribbon.

Over Time

12. Remember recent interactions.

13. Learn from user behavior.

14. Update and adapt cautiously.

15. Encourage granular feedback.

16. Convey the consequences of user actions.

17. Provide global controls.

18. Notify users about changes.

The guidelines in this group remind us that AI systems are like getting a new puppy: they are long-term investments and need careful planning so they can learn and improve over time. Learning (Guideline 13) also means that AI systems change over time. Changes need to be managed cautiously so the system doesn’t become unpredictable (Guideline 14). You can help users manage inherent consistencies in system behavior by notifying them about changes (Guideline 18).

Ideas in Excel empowers users to understand their data through high-level visual summaries, trends, and patterns. It encourages granular feedback (Guideline 15) on each suggestion by asking, “Is this helpful?”

What’s next?

If you’d like some more ideas, stay tuned for another post on this work where we share some of the uses we’ve been working with at Microsoft. We’d love to hear about your experiences with the guidelines. Please share them in comments.

Want more?

• Download and print a set of cards with the guidelines and the accompanying poster.

Authors

Mihaela Vorvoreanu is a program manager working on human-AI interaction at Microsoft Research.

Saleema Amershi is a researcher working on human-AI interaction at Microsoft Research AI.

Penny Marsh Collisson is a user research manager working on AI in Office.

With thanks to our team who developed The Guidelines for Human-AI Interaction: Saleema Amershi, Dan Weld, Mihaela Vorvoreanu, Adam Fourney, Besmira Nushi, Penny Collisson, Jina Suh, Shamsi Iqbal, Paul Bennett, Kori Inkpen, Jaime Teevan, Ruth Kikin-Gil, and Eric Horvitz.

Thanks also to Ruth Kikin-Gil for her thoughtful collaboration, and for curating examples for this post.

The race is on
The survey results show that high-growth companies are not only more than twice as likely to actively use AI compared to lower growth companies, but they also have bigger plans in a much shorter timeframe. Of the double-digit growth companies surveyed, while almost all (94%) intend to use in AI for decision making within the next three years, more than half plan to do so over the next 12 months. In comparison, the majority of low-growth companies are only looking to invest in decision-making AI in the next three to five years.

“What’s striking about the research is the difference between double-digit growth companies and those with lower growth,” says Susan Etlinger, Industry analyst with the Altimeter Group. “Double-digit growth companies are further along in their AI deployments, but also see a greater urgency in using more AI. They are looking at a one to three year timeframe – often really focused on the coming year. Lower growth companies are looking at more of a 5-year timeframe. What this says to me is that the more you know, the higher your sense of urgency is.”

Crucially, it’s not too late for those companies and leaders who are further behind in their AI journeys to start now, to increase their chances of remaining competitive.

Start small, learn fast and scale
The research findings have shown that AI is successfully utilised by leaders to invest more time in humans, while helping them create and execute new strategies. In addition, we have seen how leaders value AI’s ability to help them grow their own skills.

Evidence showing that the fastest-growing companies have invested – and will continue to invest – in AI also highlights the importance of ensuring that business, inspired by their leadership, progress on their AI journey sooner, rather than later, before they run the risk of losing their competitive edge to more progressive companies.

In Microsoft EMEA President Michel van der Bel’s words: “Start small, but start with intention. This will help teams build trust, learn from feedback and build confidence. In a nutshell, this is what will help get your AI journey off with a strong start.” Progress today, and reap the benefits for both yourself as a leader, and your company as a whole, tomorrow.

For more information on progressing your AI journey, please feel free to visit our AI business resources.
.

---

Japan’s labor-intensive fish farming sector has taken a major step toward automation with the adoption of an artificial intelligence (AI) and Internet of Things (IoT) solution that frees up highly skilled workers from a crucial, but highly time-consuming, task.

The breakthrough follows half a year of field tests at Kindai University, which plays a significant role in the national production of red sea bream – a fish known in Japanese as “Madai” that is prized by sushi and sashimi lovers both at home and abroad.

The university’s Aquaculture Research Institute hatches red sea bream and raises them to a juvenile stage, known as fingerlings. Every year, it sells around 12 million fingerlings to fish farms that grow them to adult size for the market. To meet rising demand for the delicacy, Kindai’s workers must hand sort as many as 250,000 fingerlings a day.

Japan’s aging demographics and other factors have made the recruitment of experienced sorters difficult, particularly when so much repetitive work is required. To counter this problem, the university approached its long-term partner company, Toyota Tsusho, which in turn brought in Microsoft Japan to help come up with ways of automating a number of processes. The aim is to relieve workers of manually intensive duties so that they can focus their valuable skills on more demanding tasks.

This latest innovation centers on software that automatically regulates the flow of water through pumps that transfer fingerlings from their pens and onto conveyor belts for sorting. IoT and AI tools continuously monitor and adjust the flows.

“There are three processes involved in sorting fingerlings,” explains Naoki Taniguchi, who manages the Institute’s Larval Rearing Division and is Deputy General Manager of the Aquaculture Technology and Production Center. “Firstly, we gather the fingerlings near the mouth of the pump that sucks them along with seawater from the fish pens without injuring them. To do this, we must constantly adjust the pump’s water flow to the conveyor belts. Lastly, we sort them by removing fingerlings that are too small or defective from the conveyor belts.”

Taniguchi said adjusting the water flow from a pump is crucial.

“If the flow is too fast, too many fingerlings will make it onto the conveyor belts, and our sorting teams won’t be able to keep up, and some fish that should be removed will be missed. If the flow is too slow, too few fingerlings will be sorted, and production will fall short. Until now, it’s been a process entrusted only to a few operators with sufficient experience.”

The new automated transfer system was created with Microsoft Azure Machine Learning Studio using image analysis technology that recognizes the changing ratio of fish shapes and vacant areas within a pump’s pipes. From this, the system machine-learned how expert human operators adjust flows optimally.

Field tests started last year, and within six months the system achieved the same flow control results as an operator.

Taniguchi said employees, who often used to spend their whole working day just adjusting water flows, are now able to devote their time to applying their rich experience in streamlining other fish farming processes. They will also be able to pass on their technical knowledge to a new generation of aquaculture specialists.

He hopes greater automation will make jobs in the sector more attractive to younger workers looking to build careers.

“Japan’s fishing industry employs about 150,000 people. But 80 percent of them are more than 40 years old. It is vital that we attract young people to the industry,” he said. “This automatic transfer system is just the beginning of our journey. Ultimately, we aim to automate the sorting process itself as well.”

 READ MORE:  AI and fish farming: High-tech help for a sushi and sashimi favorite in Japan

Given the accelerating pace of change around the globe, the worlds of school and work are undergoing massive transformations. New technologies such as artificial intelligence are empowering today’s students to address big challenges that motivate them, such as reversing climate change and slowing the spread of disease. At the same time, collaboration tools, mixed reality and social media are bringing them closer to one another than ever before.

To successfully navigate these changes and to leverage the opportunities ahead of them, we need to prepare students with the diverse skills they will need in the future.

To better understand how to prepare today’s kindergartners to thrive in work and life, last year we released research about the Class of 2030.

Our findings highlighted two core themes: Student-centric approaches such as personalized learning and the growing importance of social-emotional skills.

Social-emotional skills such as collaboration, empathy and creativity have long been essential, but our research revealed they have become newly important to employers and educators alike. Social-emotional skills are also necessary for well-being, which is a key predictor of academic and employment success.

So this year we decided to dig deeper, to better understand what educators and schools worldwide are doing to enhance students’ skills and well-being and to understand how technology can help. We worked with the Economist Intelligence Unit (EIU) to survey more than 760 educators in 15 countries. From Mexico to Sweden and from Indonesia to Canada, we listened to the voices of educators. We also interviewed leading experts on and reviewed 90 pieces of research.

Click the excerpt above to view the full infographic.

What we learned is that educators around the globe are placing a high priority on student well-being and they are actively seeking ways to nurture it in their classrooms, across the school environment, and in their communities.

According to the survey, 80 percent of educators believe that well-being is critical for academic success, for developing foundational literacies and for cultivating strong communication skills, and 70 percent of educators say well-being has grown in importance for K-12 students during their careers.

At the same time, school systems have not moved as quickly as educators to prioritize well-being. Only 53 percent of educators said their schools have a formal policy in place to support students’ well-being. Individual educators can do great things in their own classrooms. But to impact well-being at scale, systemic approaches are needed.

We identified some common barriers that educators encounter in trying to help improve their students’ well-being:

• 64 percent of educators said they lack the resources or time to support students’ well-being
• 71 percent of respondents said changes to enhance student well-being need to be driven by school leaders

And, we asked educators what technologies they find most beneficial in overcoming these barriers. Three areas stood out:

• 58 percent mentioned immersive experiences that allow students to explore scenarios from the perspective of others, which show strong promise for promoting social-emotional skills, particularly empathy
• 49 percent cited tools that foster collaboration among students
• 46 percent of educators favor tools that help collect and analyze data about students’ emotional states

In addition, technology provides the critical scale to take any of these approaches beyond a single classroom.

To help identify best practices, we took a close look at schools where teachers report their students enjoy higher-than-average well-being. We found several common threads. These leader schools are more likely to:

• Have a formal plan to promote well-being
• Measure and monitor well-being as well as academic achievement
• Support inclusive classroom practices that amplify student voice
• Engage purposefully with the community
• Take a whole-school approach to professional learning

A complete summary of our research results will be released in March. In the meantime, we invite you to join our free webinar series on Teaching Happiness, starting February 25, 2019, for a broader exploration of the skills that empower students to lead happy and fulfilling lives.

We are excited to be on this journey together with all of you, to learn from you, and to contribute our insights and our technologies to help every student on the planet achieve more.

Fabian Bolin was just 28-years-old when he found out he had leukemia. A promising actor, the diagnosis of cancer made him feel as if he suddenly lost control of his future and that nothing could help him regain it.

His experience is all too common.

Each year, there are an estimated 3.7 million new cases of cancer and 1.9 million deaths from the disease in Europe. According to the World Health Organization, despite making up only one eighth of the total global population, Europe bears a quarter of the world’s cancer cases. In fact, cancer is the second leading cause of death across the region behind cardiovascular disease.

While Europe is home to some of the best and most established healthcare systems in the world, cancer remains a formidable opponent. Today, leading healthcare providers and organizations are using technology such as artificial intelligence (AI) to engage and support patients, empower doctors and accelerate research. Moving us one-step closer to help manage and conquer the disease.

Giving power back to the patient
When Fabian was first diagnosed, he felt powerless and began sharing his experiences on social media. The response was so great that he helped launch WarOnCancer, a social network for cancer patients and relatives.

The original platform comprised of a 150-member strong blogging community, who represented 40 types of cancer, highlighted that most cancer patients suffer from low self-esteem and depression. With this insight, WarOnCancer is working with six partners in the pharmaceutical and broader life science industry to develop and test a new mobile app, which aims to become a global social network for cancer patients.

Scheduled to launch during 2019, the app will allow members to share their data and track how the industry uses this data in research. Through the power of Microsoft Azure, WarOnCancer can analyze this data to detect flaws and benefits experienced by different groups of patients depending on where, and how, they are treated.

“During my treatment and interactions with specialists, I was astounded to learn that almost half of clinical trials in oncology are delayed because it’s hard to find patients who meet the right criteria for that particular trial,” said Fabian. “Despite the vast majority of patients willing to share their data for clinical trials, many don’t know these are even taking place or aren’t properly informed how their data will be used. This disconnect can literally be the difference between finding a life-saving treatment or not.”

“The long-term goal is to build a ‘matchmaking’ type service for clinical trials and patients. This will increase the number of successful clinical trials, spearhead the pharmaceutical R&D-process, tailor treatment schedules and medication around a cancer patient’s needs, and ultimately save lives,” says Sebastian Hermelin, co-founder and head of WarOnCancer’s industry partnerships.

Helping doctors deliver early-detection, and increase precision and accuracy

The benefits of early cancer detection are clear. Not only does it result in a higher survival rate, but it helps minimize treatment side effects. While the process varies in every country, standard breast cancer screening typically occurs every two years and involves the mammography of women within a certain age bracket.

However, the effectiveness of mammography dramatically decreases when examining ‘dense’ breasts with a higher percentage of fibroglandular tissue. To address this challenge, the Veneto Institute of Oncology (IOV) is using a new breast density assessment tool from Volpara that has the potential to help millions of people. Leaping beyond the limits of a traditional mammogram, the cloud-based solution assesses images of a patient’s breast tissue, honing in on its density.

“Since dense breast tissue and lesions both appear white on X-rays, it is difficult to detect cancer in women with dense breasts. Moreover, it has been proven that women with dense breasts have higher risk of developing breast cancer compared to women with low breast density,” says Gisella Gennaro, Medical Physicist at the Venetian Institute of Oncology. ““But now, through advanced image analysis, we can automatically and objectively assess women’s breast density, use it to estimate their risk of developing breast cancer, and provide them with personalized imaging protocols such as using ultrasound in the event that breast density hinders cancer detection.”

“Without advanced image computing, it would be impossible to get such fast and accurate analysis. Over the next five years: we plan to examine more than 10,000 women; see an increase in cancer detection rates; a decrease in interval cancers; and sustainable screening costs. It’s truly a step forward towards precision medicine,” says Francesca Caumo, Director of Breast Radiology Department at the Venetian Institute of Oncology.

Back in Stockholm, Fabian and his team are tireless in their mission to improve the lives for everyone affected by cancer. It has been almost four years since his initial diagnosis and the journey to date has been nothing short of courageous. Alongside first-rate treatment and family support, data has also proved a somewhat hidden helping hand.

Whether its researchers, clinicians or patients – together with cloud computing and AI – humanity’s war on cancer has never been as fierce.

For more information on how Data&AI are helping clinicians, researchers and patients to make healthcare more efficient, click here.

A podcast series sponsored by Microsoft on how artificial intelligence is helping people solve previously intractable societal challenges launches Monday, Feb. 4, on This Week in Machine Learning and AI. The six-episode “AI For the Benefit of Society with Microsoft” series highlights how AI breakthroughs are advancing work in environmental sustainability, precision medicine, accessibility and life-saving humanitarian assistance.

Hosted by Sam Charrington, the podcast episodes cover technologies and people using AI to pinpoint communities that are at risk of famine before it strikes, help children with autism get additional communication tools, fight climate change through sustainable forest management and develop chatbots to efficiently connect refugees with legal services. They also explore cross-cutting themes around AI and ethics, including how to account for bias in data, ensure new technologies work for the broadest range of users and build a culture of responsible innovation.

Episodes will be available on the following dates at the This Week in Machine Learning and AI website and on Spotify, iTunes and Google Play.

• Feb. 4: AI for Humanitarian Action (podcast, transcript)
  With Justin Spelhaug, Microsoft general manager for Technology for Social Impact
• Feb. 6: AI for Accessibility
  With Wendy Chisholm, Microsoft principal accessibility architect, and AI for Accessibility grantee InnerVoice
• Feb. 8: AI for Earth
  With Lucas Joppa, Microsoft chief environmental officer, and AI for Earth grantee SilviaTerra
• Feb. 18: AI for Healthcare
  With Peter Lee, corporate vice president, Microsoft Healthcare
• Feb. 20: Human-Centered Design
  With Mira Lane, Microsoft partner director–ethics and society
• Feb. 22: Fairness in Machine Learning
  With Hanna Wallach, principal researcher at Microsoft Research

Related:

Jennifer Langston writes about Microsoft research and innovation. Follow her on Twitter.

This blog post was authored by Peter Cooper, Senior Product Manager, Microsoft IoT.

From smart factories and smart cities to virtual personal assistants and self-driving cars, artificial intelligence (AI) and the Internet of Things (IoT) are transforming how people around the world live, work, and play.

But fundamentally changing the ways people, devices, and data interact is not simple or easy work. Microsoft’s AI & IoT Insider Labs was created to help all types of organizations accelerate their digital transformation. Member organizations around the world get access to support both technology development and product commercialization, for everything from hardware design to manufacturing to building applications and turning data into insights using machine learning.

Here’s how AI & IoT Insider Labs is helping one partner, SunCulture, leverage new technology to provide solar-powered water pumping and irrigation systems for smallholder farmers in Kenya.

Affordable irrigation for all

Kenyan smallholdings face some of the most challenging growing conditions in the world. 97 percent rely on natural rainfall to support their crops and livestock—and the families that depend on them. But just 17 percent of the country’s farmland is suitable for rainfed agriculture. Electricity is unavailable in most places and diesel power is often financially out of reach, so farmers spend hours every day pumping and transporting water. This limits them to low-value crops like maize and small yields, all because they lack the resources to irrigate their crops. Additionally, irrigation technologies have an important role to play in reducing the impact agriculture has on the earth’s freshwater resources, especially in Africa.

SunCulture, a 2017 Airband Grant Fund winner, believed sustainable technology could make irrigation affordable enough that even the poorest farmers could use it without further aggravating water shortages. The company set out to build an IoT platform to support a pay-as-you-grow payment model that would make solar-powered precision irrigation financially accessible for smallholders across Kenya.

How SunCulture’s solution works

SunCulture’s RainMaker2 pump combines the energy efficiency of solar power with the effectiveness of precision irrigation, making it cheaper and easier for farmers to grow high-quality fruits and vegetables. Using the energy of the sun, the SunCulture system pulls water from any source—lake, stream, well, etc.—and pumps it directly to the farm with sprinklers and drip irrigation.

This cutting-edge solution combines ClimateSmart™ solar and lithium-ion energy storage technology with cloud-based remote monitoring and optimization software developed with support from AI & IoT Insider Labs. It’s a powerful platform that makes it simple and cheap to deploy off-grid energy and connected solutions.

Farmers get the information they need to make good irrigation decisions at scale, without the costs involved in sending agronomy experts into the field. How? SunCulture processes a steady flow of sensor data, like soil moisture, pump efficiency, solar battery storage, and other factors, that is analyzed within Microsoft Azure’s cloud environment. This sensor data is combined with data from SunCulture’s network of 2,000 hyperlocal weather stations to leverage Azure machine learning tools and provide simple, real-time, precision irrigation recommendations directly to the farmer via text messaging (SMS).

 

The platform also enables real-time locking and unlocking of devices that makes the pay-as-you-grow model feasible. The platform is smart enough to shut off pumps automatically when power levels are getting low on a cloudy day, or when optimal irrigation thresholds are reached.

How farmers are benefiting from SunCulture

SunCulture’s pay-as-you-grow revenue model allows farmers to make small, monthly payments until they own their precision sensor-based irrigation system outright, empowering even the region’s poorest smallholder farmers to take control of their environment.

On average, SunCulture customers enjoy a 300 percent increase in crop yields and a 10x increase in annual income. Farmers with livestock double their milk yield, earning an extra $3.50/day in income from milk alone. The 17 hours per week they used to spend moving water manually is now directed to better tending their crops and livestock. At a price point of $1.25/day for the RainMaker2 with ClimateSmart™, a farmer’s investment is recouped quickly, and profit starts flowing from increased agricultural productivity.

Download SunCulture’s case study to learn more.

Microsoft CTO Kevin Scott believes understanding AI in the future will help people become better citizens.

“I think to be a well-informed citizen in the 21st century, you need to know a little bit about this stuff [AI] because you want to be able to participate in the debates. You don’t want to be someone to whom AI is sort of this thing that happens to you. You want to be an active agent in the whole ecosystem,” he said.

In an interview with VentureBeat in San Francisco this week, Scott shared his thoughts on the future of AI, including facial recognition software and manufacturing automation. He also detailed why he’s “cautiously optimistic” about the ways people will devise to use intelligent machines and why he thinks Cortana doesn’t need a smart speaker to succeed.

However vital staying informed about the evolution of AI may be to the average person in the century ahead, Scott concedes it’s not an easy thing to do.

“It’s challenging, because even if you’re a person with significant technical training, even if you’re an AI practitioner, it’s sort of challenging to keep up with everything that’s going on. The landscape is evolving really rapidly,” he said.

Technologists who make and use AI today also have a duty to help people better understand what’s possible and make their work accessible, so Scott is writing a book about how AI can be a force for good for the economy in rural America.

In recent years, AI has proliferated across health care and homes, as well as governments and businesses, and its continued expansion could redefine work roles for everyone. News and public education initiatives to help citizens understand AI are important, and technologists should make their work more accessible, but Scott believes it’s not enough for businesses using AI to be disruptive in their industry.

“We have to think about how there’s balance here,” he said. “You can’t just create a bunch of tech and have it be super disruptive and not have any involvement … you have to create value in this world, and it can’t just be shareholder value.”

A ‘cautiously optimistic’ view of facial recognition

One subject that has drawn much attention from average citizens and Microsoft is facial recognition software and the potential for government overreach.

On Tuesday, the American Civil Liberties Union (ACLU) — along with a coalition of human rights and other organizations — called for major tech companies, including Microsoft, to abstain from selling facial recognition technology to governments, because doing so would inevitably lead to misuse and discrimination against religious and ethnic minority groups.

Microsoft declined to respond directly to the letter but pointed to past actions that represent its point of view. Analysis last year found facial recognition systems from Microsoft, as well as Face++ in China, were not capable of recognizing people with dark skin, particularly women of color, at the same rates as white people. Just weeks after Microsoft made improvements to the Face API’s ability to identify people with dark skin tones last summer, president Brad Smith declared that the government needs to regulate facial recognition software. Then last month the company laid out six principles it will use to govern the use of facial recognition software by its customers, including law enforcement agencies and governments, such as fairness, transparency, and accountability.

Microsoft is currently on track to implement the plan on schedule, Scott said.

Though facial recognition software could be used for nefarious purposes by businesses and governments and can drum up fears of technologically powered police states, Scott likes to think of the upside when it comes to facial recognition software use cases.

“There’s this fine line between … that boundary; there are clearly some things that you just shouldn’t allow. Like, you shouldn’t have governments using it as a mechanism of oppression. No one should be using it to discriminate illegally against people, so I think it’s a good debate to have, but I’m usually on the cautiously optimistic side of things — I actually have faith in humanity,” he said. “I believe if you give people tools, the overwhelming majority of the uses to which they will be put are positive, and so you want to encourage that and protect against the negative in a thoughtful way.”

Potential positive use cases he cites include improving security in buildings, understanding who’s in a meeting, or verifying that a person handling dangerous machinery is certified to do so.

He also offered a theoretical example based on what he observed when his wife was in the hospital last year. Just two nurses were tasked with managing an entire a hospital recovery ward, where patients were prescribed a precise regiment of ambulatory activity.

A computer vision system assigned to this task could alert nursing staff if a patient was seen in common areas too often, signaling too much activity, or if they hadn’t been seen out of their room, indicating that they were not getting enough activity.

In addition to a belief that understanding AI makes for more informed citizens, Scott emphasized that AI experts need to do more to share the positive outcomes that can come from technology like facial recognition software.

The Terminator often comes to mind in worst-case scenarios with AI, but sharing a Star Trek vision of the future is important too, Scott said, because telling positives stories helps people grasp those possibilities.

“Folks who are deeply in the AI community need to do a better job trying to paint positive pictures for folks, [but] not in a Pollyanna way, and not ignoring the unintended consequences and all the bad things that could be amplified by AI,” he said.

Scott’s book on AI in rural America

Scott believes a book will help expound on his point of view “that AI can and should be a beneficial thing for rural America.” A Microsoft spokesperson declined to share the book title or scheduled release date details.

To write the book, Scott said he began by thinking about how to define AI for his grandfather, a former appliance repairman, farmer, and boiler room mechanic during World War II.

“I think if my granddad were alive he’d be curious about AI, and part of my process is figuring out how I would explain it to him, because he wasn’t a computer scientist. And I think it’s part of your set of responsibilities these days as a tech person to try to do more of that, to make the things that you’re working on more accessible,” he said.

The book will likely draw on Scott’s experiences growing up in rural Virginia.

When asked which form of AI he believes is likely to have a more positive impact than anticipated, Scott pointed to manufacturing automation in rural areas. It’s easy to imagine advanced robotics being a disruptive factor in manufacturing, but it can also level the playing field worldwide, making it possible to establish business anywhere.

“I have talked with dozens of both small and large companies over the past couple of years, and in every last one of these conversations the thing that I’m seeing is that automation is this sort of equalizing factor, like a piece of advanced automation that runs in Shenzhen costs about the same as it does in some little rural town [in the U.S.],” he said.

“That’s this thing I think people haven’t really fully wrapped their heads around, this whole agile manufacturing movement, where you’ve got lots of these small companies that are now able to make things [and] that are repatriating jobs to the U.S. from overseas, just because they’re deploying all of this automation and their unit cost of production is dropping.”

Anheuser-Busch InBev (AB InBev) is using artificial intelligence to drive growth and innovation across all dimensions of its global brewing business.

The brewer of Budweiser, Corona, Stella Artois and more than 500 other beer brands has built a worldwide analytics platform on the Microsoft Azure cloud, enabling the company to draw data driven-insights about everything from optimal barley growing conditions to drivers of successful sales promotions.

Tassilo Festetics, AB InBev’s vice president for global solutions, shared insights about the company’s AI strategy at a recent AI in Business event in San Francisco, which Transform edited into an abbreviated Q&A.

How is Anheuser-Busch InBev using AI today?

FESTETICS:  The question is not going to be where we deploy AI, but where is it not going to be deployed, because we see it in so many different fields.

Can you share an example or two?

FESTETICS: Smart Barley, which is one of our platforms, enables us to work with farmers to use artificial intelligence to improve their yields, reduce water usage, reduce fertilizer usage and create a much more sustainable environment. We started there five years ago.

Now we see AI in the customer-facing area with chatbots and voice. Customers are expecting to have the same frictionless interaction with every company that they’re also having in their private life. Conversational bots that allow your customers to interact with your company in that way are a basic machine learning algorithm.

We also use AI in our supply chain and back-office operations. We use Azure to simplify tasks that people are performing every day and to make people’s lives much more focused on real added-value activities rather than just on transactional activities.

How did you get started in your AI transformation?

FESTETICS: When our company was born, the cloud was not there yet. Microsoft was even not there yet. We were born in 1366. So obviously we are not a digital company. We are a company that’s being digitized.

Our company has grown over time, as a large global organization our data landscape was fragmented. For us the first step was really looking at how we basically get data together, how do we harmonize it, how do we platform it. When we looked at the entire data infrastructure we said, ‘OK let’s just not touch it. Let’s hope it doesn’t break.’

We basically rebuilt everything totally as if we were starting a new company today. With advantages in technology and the cloud you can do that. And that saves a lot of time and allows you certainly to be much more agile. But for sure it’s the biggest barrier to get that data right at the beginning.

How did you develop AI expertise within AB InBev?

FESTETICS: We were very lucky that our senior management understood very early on that this was something that we should work on. So we were early to invest in new resources to join the company, because obviously we didn’t have them around. But then we also started to develop and improve the capabilities of our own people.

Last year I took my entire team to Berkeley. We spent a week on just machine learning. And it was very fun, because normally if I take my team anywhere, they are very — well, they know a lot of things. So if you put them in a room with professors after one day they will probably be explaining to the professors how life works.

In this course about machine learning you could hear a pin drop after the lesson, because everybody was still processing. And that’s, I think, the important part — that you really continue learning and you continue to build those capabilities inside of your company.

By getting new people in and by developing new skills in your people you start to see different approaches to problem solving. These people will start to find ways to deploy new technologies, new methodologies inside of the company to provide better customer service, better waste management, improved ROI on certain activities. It really starts a different way of thinking.

What advice would you give to companies that are just getting started with AI?

FESTETICS: Really start looking at your data early, because data is the fundamental part. There is no AI without data. Then start looking at the areas where you have the best business cases, where can you drive the most value for your company.

Top photo: At the Conversations in AI event, Microsoft CVP of Azure Julia White leads a panel discussion with (left to right) Tassilo Festetics, vice president, global solutions, Anheuser-Busch InBev; Abhishek Pani, senior director of AI product and data science, Adobe; Jack Brown, SVP of product and software, Arccos Golf; and Fiona Tan, SVP of customer technology, Walmart Labs.

Related:

• Watch how Anheuser-Busch InBev taps data for even better beer.
• Read more about how customers including ABInBev are using AI.
• Learn more about Microsoft’s AI tools for businesses.