Tag Archives: Natural Language Processing

Brexit, the rise of the robots and Sustainable Customer Value(s)

31 Jul

Brexit could accelerate the adoption of technology and, in turn, exacerbate unemployment. However, technology is also capable of creating equality, prosperity and a sustainable environment and value(s). First published in Campaign.

As headlines go, “Brexit leads to robot takeover” probably sounds like satire. It’s certainly up there with Brexit is “the opportunity to create a second Elizabethan Golden Age”. While both have been said in the last few days, I would like to argue that the first proclamation may actually be true.

In recessions companies are forced to make difficult choices to survive, with CAPEX and OPEX expenditure coming under ever-increasing pressure. One of the first places to look to cut costs is the wage bill. Previously the choice was off-shoring, finding a cheaper person to do the work, or the gig economy of zero hours contacts. Now as new regulation encrouches upon the fake ‘partner’ model of ‘Uberfication’, and living standards, safety and costs rise in emerging markets there is a new choice. This choice involves short-term investment, but provides long-term savings in wages, health insurance or breaks: it is the choice to adopt Machine Intelligence and automation.

Machine Intelligence, a term that encompasses Artificial IntelligenceMachine LearningNatural Language ProcessingPredictive APIs and Speech and Image Recognition, offers a potential beyond even the comparatively “basic” analytics and insight potential of Big Data.

But Machine Intelligence also means that anything with a decision tree involved can, and will be, automated. Flesh can be replaced with silicon.

While Brexit may have been partly driven by the resentment brought about by economic inequality, the recession it spawns will potentially speed-up the adoption and implementation of technologies which will accelerate unemployment. The impact of technology on jobs will far outweigh that of immigration.

Branko Milanovic’s famous “elephant graph” with China stripped out to show the impact on Western Middle Classes

Transformational technology often arrives in the form of toys, where people can play and familiarize themselves in a non-threatening way. Alternatively it targets our natural laziness and the tendency for easy-to-beat better every time. In this vein, iRobot’s Roomba vacuum cleaner has been around for so long that you can even buy one on discount next to the tills in a Robert Dyas store. While this labour saving device is not going to replace human cleaners immediately, high profile automation experiments are taking place that are more than just PR opportunities. Domino’s Pizza have produced robots that do local deliveries and Amazon are developing autonomous airborne delivery drones. These are massive companies set to make massive savings if they can strip out human costs and inefficiencies.

Universal threat

The threat to jobs is widespread, and not just to the lower paid areas of customer service or manual labour that have been traditionally impacted by technology changes. Previously “safe” professions like law are open for transformation when 160,000 parking tickets in New York and London can already be overturned by a simple chatbot lawyer.

If only 20% of marketers are trusted by their CEOs to drive growth in their business, and the average tenure of a CMO is less than 3 years, then marketing and advertising is ripe for machine disruption. Dr. Stephen Thaler claims that his Imagination Engine AI research will lead to the creation of “creativity machines” within 5 years, and we’ve already seen agencies burn out the PR-stunt of “hiring an AI creative director” while in the background machine learning is used to extract every last cent of programmatic value from media buys. The closer to the production-end of the spectrum, the more quickly it will happen.

In the world of banking and finance Citigroup predicted 1.8 million U.S. and European bank workers could lose their jobs within 10 years, as time-consuming tasks like trade capture and reporting are automated. When a bank like State Street can predict $550M in pre-tax net run-rate expense savings by the end of 2020 just through adopting digital technology then the investment case is compelling. These industry savings in human agency are being channeled into expanding Machine Intelligence’s ability to enable self-correcting, self-improvement and self-assessment. The opportunities offered by dynamic client service segmentation, contextual value targeting or A.I. correlation-based modeling, reporting and analysis have seen a huge investment in internal Fintech coder labs and external startups. But as the technology rises and replaces functions, the question is how will the “wetware” get along with the software? Machine Intelligence can already help with vetting banking clients, pricing assets, and hedging orders without human intervention, but even the personal relationship-driven deal-making of investment banking is ripe for takeover. It seems unlikely that we will see Gucci-suited bankers marching en-mass with the Occupy movement, but when they too are rendered surplus to the market’s needs then political motivations align.

It is hard to imagine wider society’s heart bleeding for advertisers and bankers, but what about the 3.5 million ordinary families supported by truckers in America alone?

The true battle for self-driving vehicles lies in enterprise and the billion tons of goods hauled every day, but at the moment this battle is being fought around the PEBSWAC (problem exists between steering wheel and chair) at the luxury end of the automotive market.

Uber is already ordering 100,000 self-driving Mercedes to replace its Uber drivers after they have, in turn, disrupted and replaced the taxi drivers. The hype is increasingly loud but tragically we have seen the first death associated with Tesla’s autopilot. The victim was a true advocate of the technology and Tesla has responded with its own statistics, claiming that with ‘1M auto deaths per year worldwide, approximately half a million people would have been saved if the Tesla autopilot was universally available’. The story of technology is punctuated with too many of these stories — few remember the first death from a Tram or his grave in a Croydon cemetery. While the true attribution of blame for the accident is still uncertain and the Rand corporation has pointed out serious flaws in self-driving car manufacturer calculations, in the correct-use cases, domains and circumstances machines ARE better than humans.

Machine Intelligence’s technologies cut across problem types (from classification and clustering to computer vision) and methods (from support-vector machines to deep-belief networks for learning). It offers the potential to discern not only the “truth” behind data, knowledge and behaviours at a scale, speed and accuracy that would be impossible to achieve without it, but also has the potential to reveal what Paul Saffo, renowned Technology Futurist and Consulting Professor at Stanford University, called the contradictions, inversions, oddities and coincidences that point the way to innovation and opportunity.

The benefits of Machine Intelligence in healthcare go beyond retrospective analysis and predictive models to influence diagnostic decision-making. IBM is currently partnering with the Memorial Sloan-Kettering Cancer Center to enable patient-specific diagnostic test and treatment recommendations for types of cancer.

Many of IBM Cognitive Computer Watson’s features that led to its famous Jeopardy victory are also relevant to the healthcare domain, including its ability to incorporate huge volumes of unstructured text data (patients’ electronic health records, medical literature, and so on), respond to natural language queries, provide probabilistic reasoning to assist clinicians in making evidence-based decisions, and improve its performance through learning from use interaction.

But what happens to the people left behind or deliberately discarded in this tornado of progress?

Benevolent social and economic solutions for a post-employment society have been suggested, including universal income or a negative income tax that at least recognises the value, social status and purpose that work provides beyond just financial means.

It is hard to be positive at the moment. It is easy to think in dystopian terms. And it is far simpler to describe the loss of existing, familiar jobs than it is to imagine the industries and functions that have yet to be created. While the role of social media manager does not even count as a blip on history’s radar before being replaced by a chatbot, its very existence in the first place is testament to technology’s ability to create jobs as well as destroy them. The challenge is what these jobs will be, and how long will the lag be between the “creative destruction” of areas of employment and the creation of new industry?

Ethical responsibilities

As a society we can be trapped in a present that is buffeted by large corporations, uber-platforms and the “wind-tunnel” pressure for hyper-efficiency and optimisation.

But now if you work in technology you also work in ethics.

It doesn’t matter if you are working on a digital campaign, service design for a new mobile purchase system or the algorithm that determines whether a self-driving vehicle chooses to allow the death of its rich sole driver instead of a group of less well-off pedestrians.

We have a responsibility. We are not politicians; we need a plan.

Technology and data must be harnessed to build something better from the debris of yesterday’s shattered dream and prevent people from being buried under it, even if there is a false temptation to blame them.

People are angry. Who to blame?

For all the talk of the central value of big data in the modern world there is a real struggle to even place a value on it. The SAS Institute in 2013 found that the market value or future income of data could not be adequately determined, and instead accounting “tricks” like including non-economic benefits or risks had to be used to tell shareholders what the value really was for the knowledge that their companies held. Machine Intelligence will enable us to extract greater value from this “new oil” and may help us usher-in a new wave of “cognitive capitalism”.

Sustainable customer value can be created through pairing Machine Intelligence with human-centred innovation. The aim should be doing something useful for people individually and as a society. We can be mindful that our work ladders-up to the Sustainable Development Goals and social balanceeconomic prosperity and a healthy environment, rather than the short termism of “the dumbest idea in the world”, Maximised Shareholder Value (MSV). Instead of following Friedman we can look to Peter Drucker’s notion that “the purpose of business is to create and keep a customer”. If Drucker is right then it is also necessary for the long term health and future of business to sustain customers — not in the sense of chasing unrealistic loyalty but in the longer term sense of ensuring there is a sustainable pool of customers to be created and acquired. The notion of Sustainable Customer Value here is co-created. It is not a self-serving fig-leaf of Corporate Social Responsibility but it changes brand and business building into an inherent dialogue about values. It is about value and values. As Paul Polman, CEO of Unilever describes it:

“I envision a 21st century form of business where the everyday consumer is helping shape the social contract … It’s a business world that is moving from value-based transactions to values-based partnerships.”

Future transactions involve interlocking feedback loops — enabling consensual coordination of JTBD and actions

While it has been stated that “the vast majority of companies struggle to tie customer experience investments to business outcomes”, the private sector can already make proven contributions to well-being as well as commerce. Maximised Shareholder Value was a successful meme in business partly due to its relatively simple measurability but Machine Intelligence, sensors and network feedback would make calculating, measuring and acting upon shared value for customers and society at scale possible. Sustainable customer value(s) could be equally simple and measurable as MSV.

This is not to argue that technology is a panacea or offers an easy, technocratic Gordian knot solution to the Wicked Problems we face. We must be mindful of the unintended consequences of seeing challenges and solutions in isolation. The car was a technological advance that created great wealth and opened up the world to previously impossible connections, but it also led to choking pollution, the emergence of car-centric cities and Walmart-style highway commerce and communities that society is struggling to move beyond. Equally placing too much faith in spreadsheets may have created a “weapon of mass destruction” or prison of financial optimization where the job of management becomes one of managing numbers and ratios, not real flesh and blood businesses that create jobs and support communities.

The spreadsheet as prison — American Beauty

Brexit itself is a classic example of The Cobra Effect and Machine Intelligence and Sustainable Customer Value(s) should not be an opportunity for top-down design or a failure to acknowledge the emergent nature of the environment, society and economy. This is not about running off and neglecting our commercial responsibilities and purpose in the pursuit of often self-reassuring Brand PURPOSE.

Machine Intelligence for our better natures

In order to succeed we would need to resolve the paradox inherent in Stewart Brand’s mantra:

“On the one hand information wants to be expensive, because it’s so valuable. The right information in the right place just changes your life. On the other hand, information wants to be free, because the cost of getting it out is getting lower and lower all the time. So you have these two fighting against each other.”

In our hyper-competitive knowledge or cognitive economy there is no incentive to share data. Companies horde and protect data. VCs and analysts place exponential valuations on the monetization of “people as data” as they “pay” for access to the latest free social platform or App. But instead of focusing on the next “collaborative economy” delivery App, could we develop a “collaborative cognitive economy” of Machine Intelligence and AI for the Common Good?

By developing a policy and infrastructure that provides for the connection of anonymized data-sets (anonymized in terms of corporations as well as citizens), the sensors powering automation and AI algorithms, we could ensure the full utilization of information beneath the surface of the economy. Essentially enshrine an AI-administered, new law of robotics that connects all Machine Intelligence beyond the manipulation of government and corporations; a law where the rules mean that entities can not only compete commercially but also ensures that the by-product is the creation of human-focused common good. It would be a world where the engines behind programmatic media are repurposed not just for bid optimisation but to create valuable personal experiences on an individual and societal level.

Progressive commuting: Algorithm anticipates train delays hrs before they occur

In this thought experiment Machine Learning could optimize bias out of the system and succeed within inherently chaotic systems given such a large training data-set and moonshot investment, with social balanceeconomic prosperity and a healthy environment the ultimate goal of collaborative nudges. It makes the final product of the internet and Machine Intelligence a benevolent (not bene-violent) ghost in the machine or All Watched Over by Machines of Loving Grace (the poem not the documentary) with encoded human compassion or consideration.

Crazy perhaps. But with the tax code and collection systems across the world already broken, it is no crazier than the European Commission’s idea to extend them to AIs as a realistic solution to the challenge of creating value that we can all benefit from, not just the machine and algorithm owners. By way of “proof of concept” Google is already using technology from DeepMind to achieve a 15% improvement in power usage efficiency. When you consider that Google used 4,402,836 MWh of electricity in 2014 (the average yearly consumption of about 366,903 US family homes), saving 15% is a huge commercial innovation as well as a sustainable innovation.

It might be easy to cry “there is no alternative (TINA)” to our current state, but the decline in Return on Net Assets and the ever-increasing gap between productivity and wages that has been happening since the late 1970s mean we need a solution to the break in late-stage capitalism: something that delivers more than just a decline in developed-economy incomes until they meet the emerging economies’ middle-class wages on the way down.

We can use brands and technology to connect positively to the people excluded from the metro-elite, the brexiters who could be left further behind, and break out of our industry bubble that lacks diversity in age, sex, race and especially class. Machine Intelligence can be the impetus in sustainable innovation rather than divergence, regression, implosion and human obsolescence. It can “expand the horizons of human creativity” and deliver Sustainable Customer Value(s).

Whatever we do for people, brands and businesses, we need to make it count.

What can brands count for and how can they deliver?
Advertisements