Connect with us


Artificial intelligence needs more than artificial trust

The technology that makes facial recognition possible is paving the way for machines to recognise feelings, writes ARTHUR GOLDSTUCK

The great irony of artificial intelligence (AI) and devices that recognise our voices, faces and fingerprints is that they are oblivious to our thoughts and feelings.

“We need to rethink our relationship with technology,” says Rana el Kaliouby, co-founder and CEO of an AI company called Affectiva. “Machines know a lot about us but are completely oblivious to our emotional and cognitive states. Yet, AI is going to change only the way we connect with our devices, but it will fundamentally change the way we connect and communicate with human beings.”

She is speaking in a packed out session at Dell Technologies World in Las Vegas, where more than 15,000 paying delegates are receiving a deep dive into topics as diverse as cloud computing and sustainability of the oceans. Her concern is that, as much as machines need to win the trust if humans, so humans must also win the trust of machines. That sounds absurd for inanimate objects, but this form of artificial trustwill be essential in a future where machines will be expected to assess both our identities and our moods, not to mention our needs.

El Kaliouby earned a PhD in machine learning at CambridgeUniversity, and helped found Affectiva in Boston, USA, to put into practice her research.

“I spent the last 20 years working to build algorithms that understand people’s emotional states, cognitive states, and apply them to the technology around us that makes them more effective.”

The reality, she discovered, is that as we imbue machines with greater intelligence, we must also imbue ourselves with a greater ethical mission.

“We need a news social contract between humans and machines. It’s a two way street. Can AI trust humans? And what will it take to have reciprocal trust? There are a lot of examples of where it goes wrong, like the chatbot on Twitter that became racist, a self driving car that kills people, and a face recognition system that discriminates against people, especially women of colour.

“Sometimes trust is explicit, but most times it is implicit, manifested in subtle interactions like tone of voice and facialexpression. The core of that is empathy. People who havehigher empathy tend to be more liable to be trusted andtherefore more persuasive and tend to be more successful in their personal lives.

“We can’t work with people we don’t trust, and I argue it is the same with AI. We have a lot of common intelligence but not enough emotional intelligence. What if a computer can tell the difference between a smile and smirk? Both involve the lower half of face but have very different meanings.”

She gives the example of the contrast between physical healthcare and mental healthcare. When people walk into doctors rooms they don’t ask what their temperature or blood pressure is, they just measure it. But in mental health care, the practitioner must ask, typically on a scale of 1 to 10, how much the person is hurting.

“The science of emotions has been around for over 200 years,since Duchenne de Boulogne mapped out stimulation of human muscles, through to a modern facial action coding system. It takes a 100 hours of training to become a professional facial analyser. It’s very time-intensive, and it’s not scalable. We use machine learning and big data and tons of computing power to speed up that process. Imagine when that becomes instant?”

The most immediate practical application of the technology is likely to be in the automotive sector, and long before self-driving cars become the norm. However, it is with cars that can switch between human-control and self-driving that the technology will come into its own.

“Our system detects four levels of drowsiness. If you are able to detect that in real time, the car can intervene in a number of ways to make it a safer driving experience. It can tell if you are using your cellphone while driving. By detecting eye gaze direction and using object detectors, the system tells us you’re not keeping your eyes on the road and looking at a smartphone. 

“How can a car react if it senses you’re distracted or drowsy? It can start with an alert. If the vehicle is semi-autonomous, it can say ‘I can be a better driver than you’, and it can take over control.

“Ina  few years, with robo-taxis, the car will still need to understand how many people are in the vehicle, what’s the mood in there, are people stressed or enjoying the ride and, if not, how can we craft the riding experience to make it more enjoyable?”

She points out that luxury car brands are in stress, because their marketing message revolves around the driving experience. Once the owner is no longer driving, the experience will still remain the key.

That, however, does not address the subtle ethical concernsthat are somewhat more nuanced than a car killing its passengers. Many supposedly cutting edge systems useCaucasian faces to “train” the algorithms to become intelligent and distinguish between faces. The result is that they have difficulty identifying non-Caucasian faces. Even within this sub-set, however, there are cultural differences that affect expressions. Affectiva addressed the issue from the start.

“We have amassed 5-billion facial frames from around world,” says El Kaliouby. “We collect spontaneous facial expressions as people go about their daily activity, and there are numerous cultural and gender differences. Women are more expressive than men but it differs by culture. So in theUK there is no significant difference, but in the USA there is a 40% difference.

“Our data is diverse, not only by gender and culture, but also context, like wearing glasses, or blurry photos, as well as by gender, age, and race diversity. It’s not perfect, but at least we are thinking about it and trying to avoid accidentally discriminating based on ethnicity.”

• Arthur Goldstuck is founder of World Wide Worx and editor-in-chief of Follow him on Twitter and Instagram on @art2gee


Now IBM’s Watson joins IoT revolution in agriculture

Global expansion of the Watson Decision Platform taps into AI, weather and IoT data to boost production

IBM has announced the global expansion of Watson Decision Platform for Agriculture, with AI technology tailored for new crops and specific regions to help feed a growing population. For the first time, IBM is providing a global agriculture solution that combines predictive technology with data from The Weather Company, an IBM Business, and IoT data to help give farmers around the world greater insights about planning, ploughing, planting, spraying and harvesting.

By 2050, the world will need to feed two billion more people without an increase in arable land [1]. IBM is combining power weather data – including historical, current and forecast data and weather prediction models from The Weather Company – with crop models to help improve yield forecast accuracy, generate value, and increase both farm production and profitability.

Roric Paulman, owner/operator of Paulman Farms in Southwest Nebraska, said: “As a farmer, the wild card is always weather. IBM overlays weather details with my own data and historical information to help me apply, verify, and make decisions. For example, our farm is in a highly restricted water basin, so the ability to better anticipate rain not only saves me money but also helps me save precious natural resources.”

New crop models include corn, wheat, soy, cotton, sorghum, barley, sugar cane and potato, with more coming soon. These models will now be available in the Africa, U.S. Canada, Mexico, and Brazil, as well as new markets across Europe and Australia.

Kristen Lauria, general manager of Watson Media and Weather Solutions at IBM, said: “These days farmers don’t just farm food, they also cultivate data – from drones flying over fields to smart irrigation systems, and IoT sensors affixed to combines, seeders, sprayers and other equipment. Most of the time, this data is left on the vine — never analysed or used to derive insights. Watson Decision Platform for Agriculture aims to change that by offering tools and solutions to help growers make more informed decisions about their crops.” 

The average farm generates an estimated 500,000 data points per day, which will grow to 4 million data points by 2036 [2]. Applying AI and analysis to aggregated field, machine and environmental data can help improve shared insights between growers and enterprises across the agriculture ecosystem. With a better view of the fields, growers can see what’s working on certain farms and share best practices with other farmers. The platform assesses data in an electronic field record to identify and communicate crop management patterns and insights. Enterprise businesses such as food companies, grain processors, or produce distributors can then work with farmers to leverage those insights. It helps track crop yield as well as the environmental, weather and plant biologic conditions that go into a good or bad yield, such as irrigation management, pest and disease risk analysis and cohort analysis for comparing similar subsets of fields.

The result isn’t just more productive farmers. Watson Decision Platform for Agriculture could help a livestock company eliminate a certain mold or fungus from feed supply grains or help identify the best crop irrigation practices for farmers to use in drought-stricken areas like California. It could help deliver the perfect French fry for a fast food chain that needs longer – not fatter – potatoes from its network of growers. Or it could help a beer distributor produce a more affordable premium beer by growing higher quality barley that meets the standard required to become malting barley.

Watson Decision Platform for Agriculture is built on IBM PAIRS Geoscope from IBM Research, which quickly processes massive, complex geospatial and time-based datasets collected by satellites, drones, aerial flights, millions of IoT sensors and weather models. It crunches large, complex data and creates insights quickly and easily so farmers and food companies can focus on growing crops for global communities.

IBM and The Weather Company help the agriculture industry find value in weather insights. IBM Research collaborates with start up Hello Tractor to integrate The Weather Company data, remote sensing data (e.g., satellite), and IoT data from tractors. IBM also works with crop nutrition leader Yara to include hyperlocal weather forecasts in its digital platform for real-time recommendations, tailored to specific fields or crops. IBM acquired The Weather Company in 2016 and has since been helping clients better understand and mitigate the cost of weather on their businesses. The global expansion of Watson Decision Platform for Agriculture is the latest innovation in IBM’s efforts to make weather a more predictable business consideration. Also just announced, Weather Signals is a new AI-based tool that merges The Weather Company data with a company’s own operations data to reveal how minor fluctuations in weather affects business.

The combination of rich weather forecast data from The Weather Company and IBM’s AI and Cloud technologies is designed to provide a unique capability, which is being leveraged by agriculture, energy and utility companies, airlines, retailers and many others to make informed business decisions.

[1] The UN Department of Economic and Social Affairs, “World Population Prospects: The 2017 Revision”

[2] Business Insider Intelligence, 2016 report:

Continue Reading


What if Amazon used AI to take on factories?

By ANTONY BOURNE, IFS Global Industry Director for Manufacturing

Amazon recently announced record profits of $3.03bn, breaking its own record for the third consecutive time. However, Amazon appears to be at a crossroads as to where it heads next. Beyond pouring additional energy into Amazon Prime, many have wondered whether the company may decide to enter an entirely new sector such as manufacturing to drive future growth, after all, it seems a logical step for the company with its finger in so many pies.

At this point, it is unclear whether Amazon would truly ‘get its hands dirty’ by manufacturing its own products on a grand scale. But what if it did? It’s worth exploring this reality. What if Amazon did decide to move into manufacturing, a sector dominated by traditional firms and one that is yet to see an explosive tech rival enter? After all, many similarly positioned tech giants have stuck to providing data analytics services or consulting to these firms rather than genuinely engaging with and analysing manufacturing techniques directly.

If Amazon did factories

If Amazon decided to take a step into manufacturing, it is likely that they could use the Echo range as a template of what AI can achieve. In recent years,Amazon gained expertise on the way to designing its Echo home speaker range that features Alexa, an artificial intelligence and IoT-based digital assistant.Amazon could replicate a similar form with the deployment of AI and Industrial IoT (IIoT) to create an autonomously-run smart manufacturing plant. Such a plant could feature IIoT sensors to enable the machinery to be run remotely and self-aware; managing external inputs and outputs such as supply deliveries and the shipping of finished goods. Just-in-time logistics would remove the need for warehousing while other machines could be placed in charge of maintenance using AI and remote access. Through this, Amazon could radically reduce the need for human labour and interaction in manufacturing as the use of AI, IIoT and data analytics will leave only the human role for monitoring and strategic evaluation. Amazon has been using autonomous robots in their logistics and distribution centres since 2017. As demonstrated with the Echo range, this technology is available now, with the full capabilities of Blockchain and 5G soon to be realised and allowing an exponentially-increased amount of data to be received, processed and communicated.

Manufacturing with knowledge

Theorising what Amazon’s manufacturing debut would look like provides a stark learning opportunity for traditional manufacturers. After all, wheneverAmazon has entered the fray in other traditional industries such as retail and logistics, the sector has never remained the same again. The key takeaway for manufacturers is that now is the time to start leveraging the sort of technologies and approaches to data management that Amazon is already doing in its current operations. When thinking about how to implement AI and new technologies in existing environments, specific end-business goals and targets must be considered, or else the end result will fail to live up to the most optimistic of expectations. As with any target and goal, the more targeted your objectives, the more competitive and transformative your results. Once specific targets and deliverables have been considered, the resources and methods of implementation must also be considered. As Amazon did with early automation of their distribution and logistics centres, manufacturers need to implement change gradually and be focused on achieving small and incremental results that will generate wider momentum and the appetite to lead more expansive changes.

In implementing newer technologies, manufacturers need to bear in mind two fundamental aspects of implementation: software and hardware solutions. Enterprise Resource Planning (ERP) software, which is increasingly bolstered by AI, will enable manufacturers to leverage the data from connected IoT devices, sensors, and automated systems from the factory floor and the wider business. ERP software will be the key to making strategic decisions and executing routine operational tasks more efficiently. This will allow manufacturers to keep on top of trends and deliver real-time forecasting and spot any potential problems before they impact the wider business.

As for the hardware, stock management drones and sensor-embedded hardware will be the eyes through which manufacturers view the impact emerging technologies bring to their operations. Unlike manual stock audits and counting, drones with AI capabilities can monitor stock intelligently around production so that operations are not disrupted or halted. Manufacturers will be able to see what is working, what is going wrong, and where there is potential for further improvement and change.

Knowledge for manufacturing

For many traditional manufacturers, they may see Amazon as a looming threat, and smart-factory technologies such as AI and Robotic Process Automation (RPA) as a far off utopia. However, 2019 presents a perfect opportunity for manufacturers themselves to really determine how the tech giants and emerging technologies will affect the industry. Technologies such as AI and IoT are available today; and the full benefits of these technologies will only deepen as they are implemented alongside the maturing of other emerging technologies such as 5G and Blockchain in the next 3-5 years. Manufacturers need to analyse the needs which these technologies can address and produce a proper plan on how to gradually implement these technologies to address specific targets and deliverables. AI-based software and hardware solutions will fundamentally revolutionise manufacturing, yet for 2019, manufacturers just have to be willing to make the first steps in modernisation.

Continue Reading


Copyright © 2019 World Wide Worx