Feb 21, 2017

IBM’s Watson focuses on eye-care advances

IBM is putting the prowess of its cognitive computing champion Watson to further use in the healthcare space, with the tech giant shedding new light on the role it can play in the treatment of eye disease.
The research, which began in 2015, involved the US heavyweight’s Melbourne-based team of researchers training Watson to recognise irregularities in eye images, which in turn offers doctors greater confidence in the early identification of patients at risk of diabetic retinopathy and glaucoma.
The technology is applied to a dataset of 88,000 retina images to analyse the key anatomies of the eye. The granular examination of each image brings to light details that may often be missed by doctors. One example of Watson’s ability is the accuracy with which it measures the separation of the optic cup and disc, a symptom of increased eye pressure, with statistical performance as high as 95 per cent.
The technology has also been trained to distinguish between left and right eye images (with up to 94 per cent accuracy), which is important for downstream analysis and effective treatment programs.
IBM Research Australia vice-president and lab director Joann Batstone said medical images were a rich source of data, which was finally being tapped comprehensively and helping medical professionals diagnose illnesses “earlier and with greater confidence”.
“We are really focused on the early detection phase but diagnosis is just one piece of the overall outcome,” Dr Batstone told The Australian.
“The transformation opportunity is actually around our ability to track and measure the progression of disease by analysing multiple images over time. It paves the way for technology to become a clinician’s assistant.”
The work done by Dr Batstone and her team isn’t the first time Watson’s ability to absorb and analyse vast quantities of data has been put to the test in a medical context. IBM is already working with Melanoma Institute Australia to help further advance the identification of melanoma using Watson and has a research agreement with Molemap, which uses advanced visual analytics to pore over more than 40,000 data sets including images and text.
Its early days but the use of Watson provides a glimpse of the impact technologies such as machine learning and Artificial Intelligence are starting to make in healthcare. ASX-listed Volpara Health Technologies’s founding director, Professor Sir John Michael Brady, a pioneer in AI and professor of Oncological Imaging at the University of Oxford, said the advancements in software was pushing the boundaries of what AI could do.
Volpara’s technology does not work quite the same way as Watson. The company’s software analyses breast density to optimise the efficacy of breast cancer screening programs and, according to Professor Brady, provides an ideal path to diagnosis.
“Error of omission is the biggest impediment to early diagnosis because it’s impossible for doctors to be on top of everything,” he said. “In that context the best way forward is to make the systems smarter.”
He said the next step for smart technologies would be to make waves in the field of cancer treatment called Radiomics — where algorithms are used to extract large amount of quantitative features from medical images — and Theranostics, which is the application of molecular imaging and genetic information of a patient to develop a personalised treatment.
Both Dr Batstone and Professor Brady are also quick to dismiss talk of AI eventually rendering clinicians obsolete.

According to Dr Batstone, getting more value out of existing resources is likely to be the main focus.

Feb 12, 2017

South Australia’s quixotic approach to energy

The free-settler state is determined to lead the nation, refusing to be enslaved by the carbon economy that dominates the former convict colonies.
“In 2015 the state government and Adelaide City Council committed to a joint aspiration,” says a South Australian government website, “for Adelaide to become the world’s first carbon neutral city.” They sure found a unique way to achieve it.
On September 28 last year during SA’s first statewide blackout, it disappeared from the view of passing spacecraft, redolent of the famous images showing North Korea wallowing in the darkness of its socialist torpor.
On that night the only emissions in SA came from the rear ends of cows and the expletive-laden curses of its citizens. The city designed by Colonel WilliamLight was in the dark and carbon neutral, very much like it was before he arrived in 1836. Except this time people were stuck in traffic snarls without signals; and in lifts.
None of this has worked out quite the way Labor Premier Jay Weatherill intended — although it is exactly how many of us predicted. Weatherill expended a fistful of taxpayers’ dollars and a truckload of carbon emissions taking a team to the Paris climate talks in 2015; he even took a video crew because there is not much point saving the planet unless someone records it.
“We are running a big internat­ional experiment right now,” Weatherill told the true believers in the City of Light, forgetting to tell his constituents back in Adelaide that he was using them as laboratory rats who could be plunged into darkness.
From websites for Renew­ablesSA and Carbon Neutral Adelaide to a range of videos, pamphlets and programs, the proselytising from the state’s Labor Party is extraordinary.
“2015 was an exciting year for our state on the topic of addressing global warming,” preaches Weatherill’s website (clueless about how exciting 2016 and 2017 would turn out to be).
The conceit is beyond parody. This is a state with a declining industrial base and the highest levels of unemployment in the nation.
It has clean air, plenty of space and, in the main, a semi-arid climate. This year its grain farmers are experiencing record crops thanks to good seasonal rains and perhaps, some farmers dare to mention, higher levels of carbon dioxide in the atmosphere.
Yet Weatherill and his team seem convinced their state can be the butterfly that flaps its wings and saves the planet. The state accounts for about 0.15 per cent of global carbon dioxide emissions.
If we shut down the state and evacuated it, the annual emissions saved would be eclipsed by China’s emissions growth within a few weeks. Indeed we could desert the state forever, leaving it to become a future Angkor Wat, and it would make not the slightest difference to global climate trends.
Yet in order to claim some kind of climate virtue, justify the odd trip to Paris and delude themselves about leading the world, SA’s policymakers have sacrificed their state’s cost of living, undermined its struggling manufacturing base and enfeebled its energy security.
South Australians have the most expensive electricity in the nation by an average of about 40 per cent, yet the power can go missing when they most need it — when especially hot or cold weather triggers high demand.
If you were doing this to actually save the planet you might grin and bear it. When you are doing it just to make an ass of yourselves it is a political crime.
To comprehend the absurdity of all this you have to trace Labor’s actions since taking power in 2002. Electricity was already a central issue because the Olsen Liberal government (for which I worked in its final two years) had privatised the system. Labor came to power promising to build an interconnector to NSW to underpin supply and boost competition. Fifteen years later it has done no such thing; rather, it is talking about the idea once again.
In the interim, under Mike Rann and then Weatherill, SA plunged headlong down the path of renewables, first achieving a 33 per cent target and then doubling down by setting a 50 per cent target (it is well on track with the renewable share more than 40 per cent already).
Mainly supplied by wind farms, the renewable investment is made under the federal government’s renewable energy target. This means the infrastructure is subsidised by other electricity suppliers, guaranteeing returns. The wind generators can often pump power into the system for extremely low prices — even for free — when the wind is blowing.
This undercut the competitiveness — as intended — of carbon-based generation. Unable to turn a profit in the new environment, gas-fired generators have chosen to close or remain idle. The state’s two main baseload generators were built in the 1960s and 80s at Port Augusta and fuelled by low-grade coal mined at Leigh Creek.
The older generator was mothballed in 2012 and the operator, Alinta, decommissioned the other last year. Before doing so, Alinta wrote to the Weatherill government seeking government support to reinvest and extend its operations in order to provide “long-term certainty” and additional retail competition.
This was rejected. On the one had this is understandable — imagine the craziness of an electricity system that subsidises renewables in order to drive out carbon-based generation but then subsidises the carbon-based plants to remain operational for reliable back-up.
But that shows the dilemma politicians have created. No doubt Weatherill, if he had his chance over, would have tried to keep the coal generators going. Certainly most of his constituents might think so.
The electricity market is extremely complex and volatile. Hence this week South Australia began load shedding when demand was high even though one major gas generator at Pelican Point was not being pressed into service.
There will be claims of profit gouging from some but alternatively, consider if the operator, French giant Engie, fires up its plant and then the wind blows. It can be stuck with gas costs but no profit.
The core of the problem is straightforward. The RET system encourages wind investment but only undermines the business model of the baseload power needed for when the wind doesn’t blow. It is the cost and complication of the renewable energy that is superfluous, not the other way around — unless you kid yourself that lower emissions in SA can change the global climate. This is an exercise in destructive futility — tilting at windmills.
The serious complication for Malcolm Turnbull is that while Weatherill’s climate crusade is all his own doing and the political consequences for him ought to be dire, it has all occurred under a federal RET that has had bipartisan support.
It is classic case of our muddled federation where we have different levels of government acting at cross-purposes. Setting a national RET at less than 25 per cent doesn’t stop self-harming states using it to achieve their own unilateral targets of 50 per cent (Queensland is aiming for 50 per cent, Victoria 40 per cent and Western Australian Labor has been flirting with 50 per cent).
The states are responsible for their own foolhardiness. And the Turnbull government’s RET ambitions seem eminently respons­ible compared to Bill Shorten’s shapeless and uncosted plan to more than double the RET to 50 per cent by 2030.
Yet Turnbull and his Environment and Energy Minister, Josh Frydenberg, need to deal with the reality that the current chaos is occurring under a RET to which they subscribe.
Remedial action is urgently needed to turn their political ascendancy on energy policy into a practical prescription.

Feb 7, 2017

Climate change: Politics and science are a toxic combination

Back in December, some American scientists began copying government climate data onto independent servers in what press reports described as an attempt to safeguard it from political interference by the Trump administration. There is to be a March for Science in April whose organisers say: “It is time for people who support scientific research and evidence-based policies to take a public stand and be counted.”
Well, today they have a chance to do just that, but against their own colleagues who stand accused of doing what they claim the Trump team has done. Devastating new testimony from John Bates, a whistleblowing senior scientist at America’s main climate agency, the National Oceanic and Atmospheric Administration, alleges that scientists themselves have been indulging in alternative facts, fake news and policy-based evidence.
Bates’s essay on the Climate Etc. website (and David Rose’s story in The Mail on Sunday) documents allegations of scientific misconduct as serious as that of the anti-vaccine campaign of Andrew Wakefield. Bates’s boss, Tom Karl, a close ally of former US president Barack Obama’s science adviser, John Holdren, published a paper in 2015, deliberately timed to influence the Paris climate jamboree. The paper was widely hailed in the media as disproving the politically inconvenient 18-year pause in global warming, whose existence had been conceded by the Intergovernmental Panel on Climate Change two years earlier.
Bates says Karl based the “pausebuster” paper on a flawed land-surface data set that had not been verified or properly archived; and on a sea-surface set that corrected reliable data from buoys with unreliable data from ship intakes, which resulted in a slightly enhanced warming trend. Science magazine is considering retracting the paper. A key congressional committee says the allegations confirm some of its suspicions.
Bates is no “denier”; he was awarded a gold medal by the US government in 2014 for his climate-data work. Having now retired he writes of “flagrant manipulation of scientific integrity guidelines and scientific publication standards”, of a “rush to time the publication of the paper to influence national and international deliberations on climate policy” and concludes: “So, in every aspect of the preparation and release of the data sets leading into (the report), we find Tom Karl’s thumb on the scale pushing for, and often insisting on, decisions that maximise warming and minimise documentation.”
This is more than just a routine scientific scandal. First, it comes as scientists have been accusing US President Donald Trump and other politicians of politicising science. Second, it potentially contaminates any claim that climate science has been producing unbiased results. Third, it embarrasses science journalists who have been chronicling the growing evidence of scientific misconduct in medicine, toxicology and psychology, but ignored the same about climate science because they approve of the cause, a habit known as noble-cause corruption.
Colleagues of Karl have been quick to dismiss the story, saying other data sets come to similar conclusions. This is to miss the point and exacerbate the problem. If the scientific establishment reacts to allegations of lack of transparency, behind-closed-door adjustments and premature release so as to influence politicians, by saying it does not matter because it gets the “right” result, they will find it harder to convince Trump he is wrong on things such as vaccines.
Besides, this is just the latest scandal to rock climate science. The biggest was climategate in 2009, which showed scientists conspiring to ostracise sceptics, delete emails, game peer review and manipulate the presentation of data, including the truncation of a tree-ring-derived graph to disguise the fact that it seemed to show recent cooling (“hide the decline”). The scientists concerned were criticised by two rather perfunctory inquiries, but have since taken to saying they were “exonerated”.
There was the case of the paper the IPCC relied upon to show that local urban warming was not distorting global data sets, which turned out to be based partly on non-existent data from 49 Chinese weather stations; the Scandinavian lake sediment core used “upside down” to imply sudden warming; the chart showing unprecedented recent warming that turned out to depend on a single larch tree in Siberia; the southern hemisphere hockey-stick chart that had been created by the omission of inconvenient data series; the Antarctic temperature trend that turned out to depend on splicing together two weather station records.
Then there was the time when a well known climate scientist, Peter Gleick, stole the identity of a member of a think tank so he could leak confidential documents along with a fake one. Stephan Lewandowsky had to retract a paper about the psychology of climate scepticism that seemed to be full of methodological flaws and bizarre reasoning.
And don’t forget Rajendra Pachauri, chairman of the IPCC for 13 years and often described as the “world’s top climate scientist”. He had to retract his “voodoo science” dismissal of a valid finding that contradicted claims from Pachauri’s own research institute about Himalayan glaciers, which had led to a lucrative grant. That scandal resulted in a highly critical report into the IPCC by several of the world’s top science academies, which recommended among other things that the IPCC chairman stand down after one term. Pachauri ignored this, kept his job and toured the world while urging others not to, before resigning over a personal scandal allegation.
I have championed science all my adult life. It is humankind’s greatest calling. That is why I deplore those who drag down its reputation by breaching its codes of conduct for political reasons, and I have no time for those excusing these enormities. They foment anti-intellectualism and play directly into the hands of people such Mr Trump. Under the Obama administration, says Professor Judith Curry, Bates’s colleague, “I suspect that it would have been very difficult for this story to get any traction.” Yikes.
Bates calls for more ethics teaching in science and for “respectful discussion of different points of view” — which we were emptily promised after climategate. It is time for the many brilliant scientists who are discovering great insights into quasars and quarks, Alzheimer’s and allergies, into neurons, fossils, telomeres and ice ages, to “take a public stand and be counted” against the politicisation of some science within their own ranks.
The Times

Feb 3, 2017

A smart home is where the bot is | McKinsey & Company

Within a decade, our living spaces will be enhanced by a host of new devices and technologies, performing a range of household functions and redefining what it means to feel at home.
The promise of devices that not only meet our household needs but anticipate them as well has been around for decades. To date, that promise remains largely unfulfilled. Advances such as the Nest thermostat by Alphabet (parent company to Google) and Amazon’s Alexa personal assistant are notable, but the home-technology market as a whole remains fragmented, and the potential for a truly smart home is still unrealized.

The rise of the homebots
The rise of the homebots
How business can prepare for the next wave of innovation.
A tipping point may be at hand. Increased computing power, advanced big data analytics, and the emergence of artificial intelligence (AI) are starting to change the way we go about our busy lives. The vision we present in this article may seem “out there,” but it simply represents the confluence of those technological developments and realization of existing trends. Those trends, along with what’s just on the horizon, according to our research, suggest to us that within a decade, many of us will live in “smart homes” that will feature an intelligent and coordinated ecosystem of software and devices, or “homebots,” which will manage and perform household tasks and even establish emotional connections with us.

A smart home will be akin to a human central nervous system. A central platform, or “brain,” will be at the core. Individual homebots of different computing power will radiate out from this platform and perform a wide variety of tasks, including supervising other bots. Homebots can be as diverse as their roles: big, small, invisible (such as the software that runs systems or products), shared, and personal. Some homebots will be companions or assistants, others wealth planners and accountants. We will have homebots as coaches, window washers, and household managers, throughout our home.

We are already entering this new era. In two years, we expect to see more items in our living space become interconnected—the formative first stage of a new home ecosystem. In five years, numerous tools and devices in the home will be affected. And in ten years, smart homes will become commonplace and will regularly feature devices and systems with independent intelligence and apparent emotion.

That level of home improvement presents significant opportunities, threats, and changes for appliances and devices that have been part of our home life for generations. The new home will be built on a foundation of platforms and ecosystems, whose producers will need to establish new levels of trust with their customers. Competition will take place not just for the consumers who inhabit the smart home, but for the interactions between consumers and homebots that increasingly will shape buying behavior. It’s not too early for a wide range of players to start laying the groundwork for success in the home of the future.

The new homebot landscape

When we envisage smart homes to come, two core features are starkly apparent.


Platforms will provide the foundation to integrate different devices while providing a consistent interface for the consumer. Frontrunners include Amazon, Apple, Google, and Samsung; start-ups at various points in the development cycle will be part of the mix, as well. The winners will deliver omnipresence through ubiquitous connectivity and go-anywhere hardware, as well as integration, with bots collaborating among each other and linking to third parties’ products and services. If the recent past is any indication, it’s likely that multiple platform standards will evolve. That will present complexities both for consumers and businesses but will foster new, niche opportunities, as well.

Product and service ecosystems

Developers will create bots that plug into the new and various platforms. In short order, this combination of platforms and bots will mature into an ecosystem of products and services. Platform companies are likely to develop their own AI-driven bots (the descendants of Amazon’s Alexa and Apple’s Siri, for example). Many other creators will develop unique homebots that integrate into different platforms, much as the apps of today have been developed for Android and iOS, which support the impressive mobile-device ecosystems we see now.

Likely, too, a hierarchy will emerge: we can expect a “master bot” that acts as general manager, juggling many services; “service bots” that handle a set of functions related to a more complex task such as managing media; and “niche bots” that perform single tasks, such as window cleaning. For now, put aside grand visions of a single, Jetsons-style Rosie the Robot replacing a human maid in toto; think instead of multiple bots performing separable, specific tasks. Well-defined scope presents much less risk of error. “If you have a robot at home,” notes Gary Marcus, a futurist and professor of psychology at New York University, “you can’t have it run into your furniture too many times. You don’t want it to put your cat in the dishwasher even once.”

Trust will be a must-have

To better understand the homebot opportunity and potential obstacles to its realization, we conducted in-home and mobile diary studies in Japan and the United States with dozens of consumers who are already using AI products or services where they live. We found that satisfaction with individual smart devices runs high. Today, people are quite willing to invite homebots into their lives to address a broad array of specific use cases: from doing individual chores to completing a more complex set of tasks to managing even certain elements of child and elder care.

But we also found there’s a crucial variable that will determine the speed and extent to which consumers truly embrace smart homes managed by homebots. The overwhelmingly determinative factor for consumer acceptance that emerged from our research was trust. Trust is initially based on the bot’s ability to perform its task, as might be expected. That does not always go as planned. But once trust is established, people are willing to cede more responsibilities to devices and systems powered by AI. One key to creating that trust will be creating bots that are more than mere automatons. After all, humans are wired for emotions. Our research confirmed that consumers are satisfied when a bot gets a task done, but they are delighted when there is a more personal, emotional element to how the bot does it.

Competing through homebots

At the same time as competitors in the smart-home space are figuring out how to create trust, they also must learn how to compete in a new landscape where the winners are influencing the homebots themselves. As consumer–bot interactions become a new nexus of competition, a variety of players will need new skills in designing bots, marketing products and services to them, and building business models that exploit their position at the center of the home.

Designing bots

Increasingly, designers will tap into and even advance data science to develop solutions that go beyond addressing static insights. Likely, that will entail solutions that are at least in part AI-driven, in order to react instantly and evolve constantly for the needs of customers. By understanding customers through a variety of approaches including ethnographic research and AI-generated insights, designers can help guide businesses through the complicated tangle of interactions and diverse engagement models. We expect solutions will migrate from screen-dominated interfaces to more physical and even atmospheric interactions. Companies that have more compelling and intuitive engagement models between bots and consumers—and can achieve significant market penetration first—will hold the competitive advantage.

To become machines that are truly integral to peoples’ home lives and to establish genuine trust, bots will need to connect with and relate to humans. That’s hard, and it goes beyond AI to the realms of artificial emotion (AE). AE encompasses attributes such as tone, attitude, and gestures that communicate feelings and build an emotional connection. Consider Alexa. Several of our interview subjects told us that they think of Alexa as a friend. That doesn’t develop from merely providing the train schedule when asked. It comes because Alexa evokes a sense of support, through its sensitive omnipresence and nuanced voice interaction. Interacting with Alexa really is like talking to a friend.

Marketing products and services to bots

As consumers trust bots more and in turn cede to bots more control over their home management, people will become less involved in the active decision making that goes on in daily home life. For providers of home goods and services, this means that bots will increasingly become the customer— or at least an important intermediary between a selling business and a human purchaser.

Marketing for bots certainly gives new meaning to the term robocalls. But it also poses a serious challenge: How can businesses position their products and services to a bot so the human consumer will passively allow, or actively ensure, a purchase (exhibit)? We expect that the marketer’s mission will be comparable to the steps one takes to rank one’s product or service at the top of an Internet search result. Just as companies focus on search-engine optimization, they will need to develop metadata and tagging systems that are optimized for homebots.

Marketing to consumers will increasingly mean marketing to their bots as well.
Given the simplicity of automated purchases and refills for many household products, sellers will need to focus on getting into a homebot’s “consideration set” and optimize features to win the likely comparisons embedded in a purchase-decision algorithm. That calls for an approach that is much harder than “one and done.” Given the speed and reach of AI, providers will have to monitor bot purchasing behaviors continuously and be vigilant in tracking competitors’ moves going forward.

The stakes are real; a shift in AI preference toward a competing product could reduce demand to zero. The once all-powerful intangible power of a brand may now be reduced to a tangible sum of its parts. As AI gathers inputs across consumer networks, unpleasant consumer experiences or negative feedback could have near immediate impact on bot purchasing preferences. As a result, analytics and marketing will need to be rapid, responsive, and agile. Consumers who can’t be bothered to search for the right purchase or are overwhelmed by the complexity of choice can have a homebot scan constantly based on variable individual preferences (such as cost, appearance, and durability).

Evolving business models

We expect that a wide range of homebot business models and use cases will emerge. Not only could homebots be purchased or rented for a specific task, people may share or rent them out to others. It’s conceivable that networked bots will work together across households, for example, to increase processing power, share expenses, or even partake in buyer co-ops to benefit from bulk pricing. Each of these models creates opportunities for new revenue streams.

The greatest source of value may come from the data. Bots will acquire and generate reams of information, and these data points will be critical for increasingly data-driven projects and services. Data will be sources of insight and even products in their own right. And understanding the implications, opportunities, and information about the smart home won’t be someone’s part-time job. It will require a dedicated team to parse the data, develop strategies, manage partnerships, and drive experiments that will become integral to creating value.

Laying the foundation

Businesses that seek to compete in the smart home can begin their housework early. A network of functioning bots is, in effect, an ecosystem of capabilities. Each bot will need to follow standard protocols to communicate with one another. But while a house may be bounded by four walls, a homebot ecosystem extends into the ether; it has to, as bots will need to interact with markets and networks around the world. Smart cars, wearables, and mobile devices are but a few examples. How all those systems “talk” to one another will be the core IT challenge for the foreseeable future.

On the technical side, mastery demands an intimate understanding of AI technologies and how they work with one another. On the strategic front, it’s worth the effort to identify what your company’s competitive advantages are or may become and then imagine how these advantages could align with the homebot value opportunities that are likely to emerge. Remember: the smart home will require different parties to work together. It’s not too soon to take note of players developing complementary—or potentially competitive—capabilities, and consider opportunities for potential partnerships. Most important, keep in mind that the success of homebots and smart homes is not wholly about technology. Rather, smart homes and bots are about how technology makes us feel. The objective is to meet the needs of human consumers and to make a house feel like home.