The Weird CEO

by Charles Towers-Clark

  This is particularly important when it comes to assessing the quality of the data used. In 2016, Microsoft launched an Artificial Intelligence ‘teenage girl’ bot. Once released onto a cynical world, it was hounded with racist speech, which it interpreted as normal and thus started spewing propaganda.[lxi]

  Data bias can have a damaging impact beyond teenage girl bots. In the US, a confidential algorithm called COMPAS is used to predict the likelihood of re-offending by criminals[lxii] as well as to guide the sentencing process. According to analysis undertaken by Pulitzer Prize-winning non-profit news organisation ProPublica, the algorithm had an inaccurate bias as black people re-offend less than the algorithm’s prediction and white people re-offend more.

  The problem is that the data used to build these algorithms and provide assumptions have a human bias.

  It is possible that an algorithm could be programmed to try to overcome this human bias, although this could lead to even greater confusion when trying to understand how the algorithm arrived at a decision. However, if developers do not take account of human bias in the data sets, it will become more and more difficult to fight discrimination as Deep Learning algorithms build upon themselves with flawed data.

  Whilst Deep Learning is potentially scary while being full of promise – let’s take a reality check. Only 1% of Artificial Intelligence is Machine Learning and only 1% of Machine Learning is Deep Learning.[lxiii]

  E)

  THE EFFECT OF AUTOMATION

  “Automation is good, so long as you know exactly where to put the machine.”

  Eliyahu Goldratt – Author

  The good news is that any mass loss of jobs won’t happen overnight. What some may consider to be bad news is that it has already started.[lxiv]

  We are already about 5,000 years into the process, but automation has accelerated significantly in the last 50 years, and even more in the last 5 years. As we start to automate, further automation becomes easier and hence the speed is increasing at an exponential rate.

  One of the key indicators of this is the speed of payback for robots. In China this had already dropped from five to three years between 2013 and 2017.[lxv] As prices continue to drop, owners are more likely to rely on a computer, that comes with a guarantee and can be easily replaced, than a person that needs training, could leave or fall ill.

  So, what will drive this ever-faster uptake of automation?

  Firstly, the speed of innovation. Futurist Ray Kurzweil created a formula to work out the total calculations per second (cps) that a human brain can undertake so that this could be compared to a computer.[lxvi] He calculated it at 10 quadrillion cps (in case you are wondering, that is 10 followed by 15 zeros – so quite quick). There is one computer that beats a human brain – built in China with a speed of 34 quadrillion cps at a cost of $390 million and using 24 megawatts of power (the output of a small power station compared to the human brain that needs the same power as a faint light bulb). The cps of a $1,000 computer is about one thousandth that of a human brain but it is worth remembering that, in 2005, the same computer had a millionth; in 1995 a billionth; and in 1985 a trillionth of the speed of the human brain. When will the cps of a $1000 computer surpass a human brain? Probably somewhere around 2025.[lxvii]

  Secondly, more and more investment is being directed towards Artificial Intelligence and robotics. IDC forecasts that worldwide spending on AI hardware, software and services will jump from $12 billion in 2017 to $58 billion by 2021.[lxviii] Just as venture capitalists in 2000 wanted to invest in a dot com, many now expect an element of Artificial Intelligence in any investment opportunity.

  Thirdly is legislation. As mentioned previously, governments have created little legislation around Artificial Intelligence but, when they do, automation will accelerate. It can feel to those outside Silicon Valley or Silicon Fen that corporates are driving innovation policy as governments are slow in embracing new technology.[lxix]

  An exception is Estonia, where some innovators prefer to work for the government than for private companies. Estonia has implemented a digital society where almost everything can be done online. Not only does this apply to any interaction with government institutions, but private companies can also interact with user information.

  The Estonian government did not try to layer an online environment on top of an existing bureaucratic process, but rather built a digital environment that could be layered upon. One of the key attributes of the system is that no data have to be entered twice, so there is coordination (called X-Road) between multiple databases holding the same information. This makes it easier to provide privacy controls to each user (and citizen) who can see all the information held about them and which organisations have been given permission to access each piece of information. Furthermore, whenever data are requested or viewed by a third party, that enquiry is logged with a record of the viewer. Alerts are created if information is requested that is not deemed necessary to the viewer, and a phone call may be made to ask why the information was accessed.[lxx]

  The UK has been trying for years to digitalise medical records within the National Health System[lxxi] but is nowhere near the level achieved in Estonia, where not only is information shared within hospitals and available to emergency services, but is automated sufficiently that prescriptions are available at private pharmacists by presenting an ID card and all appropriate discounts are calculated automatically.

  Deaths by accident have dropped by 50% in Estonia over the last 20 years[lxxii] due to the speed of response by emergency services and to preventive measures. Emergency services can identify the location of an accident victim to within five meters and 93% of calls are answered within ten seconds.

  Obviously, this level of digitalisation has led to a loss of jobs but Estonia has made a huge effort to retrain – and now has one of the highest rates of technology employment in Europe. Within such an environment of innovation, high technology employment isn’t perhaps a huge surprise – after all Skype was created in Estonia.

  Estonia has one major threat. Sharing its eastern border, Russia is keen to extend its influence over the country. As part of NATO, Estonia is a strategically important geopolitical country, and the threat of invasion from Russia is real. It is for this reason (as well as cost) that the Estonian government holds its citizen’s information in the Cloud (hosted in a second site in Luxembourg) so that the government could operate virtually if required. Furthermore, by digitalising its citizens’ records, Estonia is reckoned to save about 2% of GDP per year, which corresponds to the amount that it spends on defence – mostly to guard against Russia.[lxxiii]

  This digitalisation has also changed how Estonian society interacts with itself. A level of trust has been introduced both towards the government and towards private enterprise. Because citizens feel that they own their data, an open and more transparent society has been created.

  There is no reason (apart from will and politics) why larger countries cannot follow the Estonian model.

  Estonia is also one of the first countries to legalise self-driving cars (to date, Level 3 – a human being needs to be in the car).[lxxiv] Other countries (including the US and the UK) are moving forward with trials and legislation[lxxv] so it is now a question of when, not if, they become legal. Within five to fifteen years, there will be more self-driving than manually driven cars[lxxvi] and the number of conventional vehicles will disappear very quickly as they become uneconomical. Therefore, in the decade following self-driving legislation, almost all jobs that require driving skills will be lost.

  A study by the World Bank predicted that, globally, 57% of jobs will be lost to automation within the next 20 years.[lxxvii] This will affect developing countries with repetitive, low paid jobs to a greater degree than developed countries.

  However, as mentioned earlier, within developed countries some jobs will be affected by automation to a greater degree than others. In 2013, Carl Frey and Michael Osborne wrote a paper on those professions most likely t
o be automated. They may have underestimated the speed of automation in certain industries but their point is valid. Of the 702 professions, 25% have a 90% or above chance of being automated[lxxviii] In terms of timeframe, again five to fifteen years would be a good estimate.

  There are plenty of studies about job losses in the future. However, there is a better example in the present. By 2018, Amazon had about 100,000 robots. For each robot employed, approximately two jobs are destroyed within companies that compete with Amazon.

  The natural assumption is that low skilled jobs will be lost first. However, David Autor from MIT came to a different conclusion.[lxxix] He pointed out that US employment statistics could be read to suggest that both middle and highly skilled jobs are being steadily replaced by machines. As a higher number of low skilled jobs are being created in the workplace, it creates an illusion of keeping employment numbers high.

  The three previous industrial revolutions each caused a loss of manual jobs. There is a good possibility that the fourth one will destroy more office-based than manual jobs. The reason for this can be explained by looking again at IBM Watson, which is trying to replicate human cognitive behaviour. It may eventually match a human brain (although that is debatable) but it will always be limited by a lack of physical attributes – which is why manual jobs may last longer than office jobs.

  The potential for automation therefore is not the skill level, but rather the replicability of the job. The more replicable it is – the more likely it is to be lost.

  Conversely, the types of jobs that are difficult to replicate are those that require:

  1) Creativity (Scientist, Business Strategist)

  2) Human relations (HR, Business Development)

  3) Management of unpredictable tasks (CEO)

  F)

  THE DANGER OF SPECIALISING

  “I can probably earn more in an hour of writing or even teaching than I could save in a whole week of cooking. Specialization is undeniably a powerful social and economic force. And yet it is also debilitating. It breeds helplessness, dependence, and ignorance and, eventually, it undermines any sense of responsibility.”

  Michael Pollan – Author

  Our Innovation Director was adamant that developers could not be replaced by computers, but a CEO could. I responded in the same way as most employees: “My job can’t be replaced by a computer!” Whether I was correct or not, acting as an ostrich and assuming that a job cannot be replaced by a computer will lead slowly but surely to a loss of that job – to the competition. Somewhere, either around the corner or across the world, another company will be working out how to automate tasks that will make their product cheaper and probably better than yours.

  Each modern-day job consists of a series of tasks – some of which can be more easily replicated by computers than others. This is summarised in the table below.

  Table 2.2 The Effect of AI on Office Jobs

  Office Job

  Effect from Artificial Intelligence

  Computer Programmer

  Computers are already starting to write code that will replace programmers. By devising AI programs computer programmers will write themselves out of a job. However, programmers who understand business requirements will still be required to write test code.

  Customer Support Agent

  Voice recognition, continuously learning knowledge bases and automated voice responses (from knowledge bases) will remove the need for customer support agents.

  Bookkeeper and Accountant

  Most finance tasks can be processed and therefore will be automated. Senior management accountants will be needed to bring multiple sources of information together and combine this with Emotional Intelligence.

  Lawyer

  Already computers can analyse and respond on case law. Smart contracts and blockchain will remove much need for lawyers.

  Business Developer

  Difficult to replace as each opportunity can be unique and relates across multiple departments. Also uses multiple sources of information, which is costly to program into computers.

  Sales Person and Purchaser

  Much of sales and purchasing has been and will be more automated using purchasing platforms. Amazon has already changed the future of selling.

  Marketeer

  Too many different roles to be able to generalise. Marketing initiative will be required to try new methods and ideas. Other jobs which can be processed (including content writing or lead generation) will be automated.

  Operations Manager

  The role of operations is to create and fulfil processes – these should be automated. Operations manager will be required only to automate new processes

  Humans Resources Manager

  HR processes (payroll etc.) will be automated. A good HR Director focused on building or maintaining the company culture will have value.

  CEO

  CEO has to deal with a lot of different tasks, including pushing people to keep momentum. AI will help with strategic decision-making. CEO should aim to become redundant by pushing ownership and responsibility down the organisation.

  3. BUSINESS CHANGE

  “The underlying source of anguish for many people in work today is an antiquated system of employment and management designed for an industrial age.”

  Richard Donkin – Author

  The need for business change is simple – companies must survive and thrive in the face of ongoing competition, which involves handling technological changes as well as attracting and motivating the best staff. Today’s workforce expects a different working experience than that provided by many hierarchical organisations and self-management provides a preferable environment.

  A)

  SELF-MANAGEMENT

  “Humility is what makes teams great.”

  Rick Pitino – Basketball coach

  As outlined in Chapter 1, most organisations are dictatorships and their long-term future is dependent upon the strategic decision-making abilities of the CEO. When companies do not thrive, despite generating revenue, their failure is usually caused by the CEO not understanding, or not responding to, the need for change. I have certainly been slow to react to strategic necessity in the past and I now realise that, for an organisation to be successful, it must not depend upon the CEO alone but, rather, each and every person should provide initiative.

  This can be facilitated within self-management. Whilst every organisation has its own culture and personalities and is therefore unique, a key part of any flat organisation is smaller teams which avoid inertia and the ‘not my job’ attitude found in large groups.

  Deciding the size of a team will depend upon three questions:

  What is the objective of the team?

  What skills are needed to achieve that objective?

  Which and how many people are needed to cover those skills?

  The way in which teams are created and organised depends largely upon the dynamics of each company. The methodology that we have started to use within Pod Group is covered in Chapter 6.

  Having created more manageable groups, flat organisations need varying levels of reporting and decision-making – generally depending upon the level of self-management that teams are given. Let’s take each of these in turn.

  Reporting can be done for informational or control reasons. If a team is reporting to one or a series of bosses, in positions higher than all the individuals in the team, it will probably be used as battering rams for the bosses to question how the group is working. If, however, the reports are disseminated across, not up and down, the organisation, especially to parallel teams who need the information to complete their work, then they can be valuable.

  The way that decisions are made can be similar to and linked to the reporting process (especially in a hierarchical organisation). Alternatively, decision-making can be a separate process that takes place within a team but does not relate to the reporting of activities to other teams.

  A number of companies around the world
have embraced the concept of totally self-managing teams, which fulfil all the functions of a normal hierarchical organisation. WL Gore, started in 1958, has 9,000 employees across thirty countries. Known for its Gore-Tex products, it has evolved into many other businesses. Its founder, Bill Gore, when he started the company, envisaged a flat structure with autonomous teams making all the decisions required to run their unit. WL Gore uses many processes to stay successful, but one of the most fundamental is that no part of the organisation can grow beyond 200 people.

  Self-managed teams are more about the approach and thought process behind work than the structure; and herein lies the problem – it can seem very ambiguous. However, a number of aspects are common amongst all self-managed teams. The first and most important is that bosses need to allow their employees make their own decisions.

  As mentioned earlier, this requires a basic assumption that people are good and, if trusted to do so, will generally try and do their best.

  Not only does the CEO need to let others make decisions, middle managers also need to let go and this is often even harder to achieve. When their role is usurped, the middle managers’ existence is called into question. Their value in the team then comes not from their position, per se, but from what they can add. This explains why so many managers leave companies after they become self-managed.

  By removing the traditional managerial role, a level of self-motivation is required in each participant of the team to bring out the best in themselves and others. There are no orders to follow, so experience and initiative are required to work out what needs to be done. Training can be an important aspect of self-managed teams where individuals are taught not only the functional aspects of their jobs, but also the soft skills required to work with others. As a result, members of the team are likely to be more motivated and responsible for the work that they are doing.