AI offers the impression that the creation of new jobs and opportunities will account for the temporary loss of certain replaceable jobs. However there is also a strong argument that new jobs will be concentrated in certain parts of the developed world, whilst posing a threat to developing regions such as Africa through the substitution of low-wage jobs with robots that could ultimately displace millions of jobs in the future. An example of this trend is automated systems, substituting many semi-skilled jobs present throughout administrative, labour or customer relations roles in automotive, banking, health, insurance, accounting and call centre industries which are typically outsourced to developing countries such as India, Vietnam, south Africa and Morocco. Poorer nations have used their relatively low-cost labour force as a driving force to catch up with developed economies, examples being China, Thailand and Vietnam. This method of outsourcing may soon come to a close as the growing influence of AI and supplementary technologies like 3D printing will encourage consumers to manufacture locally with the convenience of customisation. The IT industry also encourages skilled workers to migrate to safe and developed cities that offers career opportunities like Silicon Valley, but exacerbates skilled labour in developing cities, further driving inequality between regions. It is critical that AI is used to serve the broader society such as boosting productivity in developing economies and enabling greater access to education, health, employment and other opportunities. M-Pesa and M-Tiba serves as a great example by facilitating mobile phone-based money transfer and health services in Kenya. It is crucial for policy makers to confront the rapidly and far reaching impacts of AI and make decisions that will benefit society. Photography was a chemistry issue As Arithmetic became cheap, it was replaced. This is now in Banking, music, communication etc. CES (robotics, drones, VA, Economists say AI is in a different category. The rise of AI is the drop in cost of prediction. Prediction: taking information you have to generate information you don’t have. e.g: Demand forecasting, classification. The cost of prediction is plummeting, and the implication is that we will be using significantly more of it (better faster cheaper), we will start converting non prediction problems into prediction problems (autonomous driving was once seen to be impossible because it would require programming of an infinite number of variables in an uncontrolled environment. This has since changed with AI which uses predictions of a safe driver’s actions. Language translation was once rule based issue requiring expensive resources such as linguists, but is now a prediction machine. Google will soon launch a commercial translation device called google pixel buds, which uses AI). Input, Prediction, judgement( how costly would it be for the prediction error), action, Outcome, Feedback. Implications for jobs in the economy. The substitute machine intelligence (Value of human prediction falls as the capability of machine prediction increases and its cost decreases. Human prediction is very flawed, it has systematic biases and is very slow and leads to failure often and thus future businesses will be likely to incorporate machine prediction over human labour, a bleak view of the future. However the value of other factors like Input, Judgement, Action, Outcome and feedback will increase as a complement of Prediction. This means that data will become more valuable as an aspect of input, since it can be used with variety as a result of cheaper prediction. Our judgement increases dramatically as the value of machine prediction increases, a doctor can They’re the things that increase in value; data is the new oil AI as a prediction tool is actively increasing its accuracy and quality of prediction at a dynamic pace, and could have far reaching impacts that could surmount to a Amazon is making efforts to improve their prediction accuracy of consumer choices and with enough progress, this could fundamentally disrupt their business strategy of customer shopping and shipping. Amazon’s patented and tested anticipatory shipping strategy through clothes in certain markets, serves as an example of this future disruption as Mobile to AI strategy Science of how we see Visual Analysis visual perception 1. We don’t go in order Visuals aren’t read in a predictable, linear way, as text is. Instead, we look first at the visual and then scan the chart for contextual clues about what is important. We should write charts spatially. 2. We see first what stands out Our eyes go directly to change and difference, such as unique colours, steep curves, clusters or outliers. 3. We see only a few things at once The more data that’s plotted in a chart, the more singular the idea it conveys. If we need to focus on individual data points, we should plot as few as possible so that the visuals don’t disappear into an aggregate view. 4. We seek meaning and make connections Our minds incessantly try to assign meaning to a visual and make causal connections between the elements presented, regardless of whether any real connections exist. 5. We rely on conventions and metaphors We use learned shortcuts to assign meaning to visual cues based on common expectations. For example, green is good and red is bad The Two Questions 1. Is the information conceptual or data-driven? 2. Am I declaring something or exploring something? 1. Conceptual or Data Driven? Conceptual: Focus in Ideas, and the aim is to simplify, and teach - E.G (Here’s how our organisation is structured) Data Driven: Focus in Statistics, and the aim in informing, enlightening - E.G (Here are our revenues for the past 2 years) - Data driven charts could take on a conceptual form and vice versa 2. Declarative or Exploratory? Declarative / Confirmatory : Focus in documenting, designing and the aim is in affirming the message - Declarative E.G(Here are our revenues over the past five years) - Confirmatory; is what I suspect actually true? And what are some other ways of looking at this idea? E.G ( Lets see if seasonal cycles are the causes to a dip in sales) Exploratory: Focus in prototyping, iterating, interacting, automating - Corroborate, E.G(Let’s see if marketing investments contributed to rising profits) - Discover, E.G (What would we see if we visualised customer purchases by gender location?) Declarative Idea Illustration Everyday dataviz Data/info type: Process, framework Viz type: Simple, metaphorical Viz features: Convention, metaphor Typical setting: Presentations, teaching Talent skew: Design, editing Goals: Learning, simplifying Examples: process diagram, cycles, trees, bridges, hierarchies, pyramids. Data/info type: Simple, low volume Viz type: Conventional chart, static Viz features: Clear point, simple narrative Typical setting: Formal, presentations Talent skew: Design, storytelling Goals: Affirm, set context Examples: Line/bar chart that gets the message across in the right context Data Driven Conceptual Visual Discovery Idea Generation Data/info type: Complex, undefined Viz type: Creative, metaphorical Viz features: Convention, metaphor Typical setting: Working session, Brainstorming Talent skew: Team Building, facilitation Goals: Discovery, simplification Examples: Informal settings, notes/scribbles Data/info type: Big Data, complex, dynamic Viz type: Advanced, unconventional Viz features: Interactive, autodynamics Typical setting: Working session, testing, analysis Talent skew: BI, programming Goals: Trend Spotting, sense making, deep analysis Examples: Complex network diagrams, rough scatterplots, Function over form Exploratory Data Visualisation Process: Sketch 1. Prep: basics for reminders, material prep, intended audience and context 2. Talk and listen: Discuss with associates, phrases, words, statements that captures idea 3. Sketch: Match keywords to chart, try different visual approaches 4. Prototype: more accurate detailed sketch, Refine 1. Focus on design structure and hierarchy Title, subtitle, visual field, source line, align elements, consistent weights 2. Charts that are straightforward should focus on design and clarity Remove unnecessary elements, make sure all elements support the idea 3. Design simplicity leads to elegance Show only what’s needed, avoid belt and suspenders, limit eye travel. Refine to persuade 1. Hone the main idea This will expose where and how you can focus your energy on persuading an audience What am I trying to say or show? I need to convince them that … etc 2. Make it stand out Use simple design to support the main idea 3. - Emphasize: the main idea by adding Isolate the main idea by reducing unique attributes Adjust what’s around it Remove reference points Add reference points Shift reference points Persuasion vs Manipulation 1. The truncated Y Axis Chart that removes value ranges from y-axis, thereby removing data from the visual field. Advantages: emphasises changes, makes curves curvier and distance from points bigger Disadvantages: It can exaggerate or misrepresent change, making increases look steep. 2. The double Y Axis Chart that includes 2 vertical scales for different data Advantages: compels viewer to make comparison between datasets Disadvantages: relationships between different values are artificial. 3. The Map Geographical boundaries to encode values related to that location, such as voting results by region. Advantages: Geography is a convention that allows to find data quickly on basis of location. Disadvantages: Size of region doesn’t necessarily reflect the data encoded. Present to persuade 1. Show the chart and stop talking 2. Don’t read the picture 3. For unusual visual forms, guide the audience 4. Use reference charts 5. Ged rid of chart if you something important to say 6. Show simplicity but include more information Presentation engagement tips 1. Create tension: show part before the full visual, ask audiences to speculate 2. Use time: Reveal large values gradually for easier understanding 3. Zoom in or out: Increase or decrease scale to show value to understand 4. Bait and Switch: Lure viewers in with a visual they may expect to see and then show actual version 5. Deconstruct and reconstruct: Break down visual into smaller parts 6. Tell stories: Use dramatic structure of setup, conflict and resolution to make charts tell a story Visual Criticism 1. Make a note of the first few things you see 2. 3. 4. 5. Make a note of the first idea that forms in your mind and then search for more Make notes on likes, dislikes and wish I saws Find three things you’d change briefly Sketch and prototype your own version, and critique yourself Comparison Composition Variable width chart Comparison among items 2 variables/item Stacked 100% Bar chart Changing over time Only relative differences matter Few periods Table with embedded charts Comparison among items Many categories One variable/item Stacked area chart Many periods Relative and absolute differences matter Many Periods Pie chart Static Simple share of total Bar Chart Comparison among items & Time One variable and few categories Circular Area chart Comparison over time Cyclical data Line chart Comparison over time Non-cyclical data Many categories Relationship Waterfall chart Static Accumulation or subtraction of total Tree Map Static Accumulation to total and absolute difference matters Distribution Scatter Plot 2 variables Histogram Single variable Few datapoints Scatter Plot bubble size 3 or more variables Line histogram Single variable Many data points Scatter plot 2 variables Alluvial chart Maps Network Logic Flow Chart Geography Hierarchies 2X2 Networks In the case of salon reservations, customers were booking time with a stylist whereas restaurants were booking seats Simplicity, development board meetings Disruption to the business strategy can lie in AI. Since AI are prediction machines, as the level of prediction accuracy increases, Recruit could potentially override all verticals of its business where it would begin to book customer’s sessions before their intention. 1. 2. 3. 4. 5. Adecco Group Randstad Manpower Group Allegis Group Recruit Holdings Co., Ltd is the world’s 4th largest staffing company Recruit Holding’s Current 46 Billion dollar valuation - Permanent and temporary staffing HR services Executive search and online recruitment Provision of advertisements an value added services in travel real estate automobiles, dining, beauty, and weddings Goals 1. Largest staffing agency in the world in terms of number of positions filled by 2020 2. Largest marketing media company by 2030 3. Recently announced that they want to have the most users by 2030 Strategy 1. Strengthening operations in Japan 2. Expanding globally through mergers and acquisitions 3. Creating new businesses by leveraging data analytics, machine learning and AI Leaders - Ken Asano CEO of Recruit Lifestyle Yoshihiro Kitamura managing Corporate executive officer of Media and Solutions Alon Halevy CEO of Recruit Institute of Technology (RIT) History Recruit was rocked by a major political bribery and insider trading scandal which forced the retirement of the Prime Minister Noboru Takeshita as well as Recruit’s founder Hiroshima Ezoe. Coupled with this scandal and the sharp downturn in the financial markets and the real estate industry, Recruit’s growth was stunted and was subsequently ¥1.4 trillion in debt. Since this drastic event, Recruit rapidly tightened its ethical rules recovering the trust of consumers and clients and took advantage of the rapid emergence of the internet in the 1990s, identifying new key markets in house rentals, bridal, travel and hospitality to provide information services. Today the sustained level of Recruit’s ethical responsibility is evident where their primary goal, avoids promising certain sales or financial figures as it may pressure the organisation to partake in unethical actions. It however maintains strong ambition with its goals of reaching the most users by 2030. As recruit’s operations begins to be more involved with AI and it’s capacity to replace rudimentary roles, it is crucial to be aware of its potential in driving inequality across other regions, and displacement of jobs. Recruit’s transition from a paper magazine medium to the internet has in part reduced the demand for printing services but also increased the need for jobs in cybersecurity, webdesign etc. AI can allow recruit to access developing country markets with a superior unmatched service and uplift many semi skilled labour in the process. The IT industry also encourages skilled workers to migrate to safe and developed cities that offers career opportunities like Silicon Valley, but exacerbates skilled labour in developing cities, further driving inequality between regions. AI offers the impression that the creation of new jobs and opportunities will account for the temporary loss of certain replaceable jobs. However there is also a strong argument that new jobs will be concentrated in certain parts of the developed world, whilst posing a threat to developing regions such as Africa through the substitution of low-wage jobs with robots and AI that could ultimately displace millions of jobs in the future An example of this trend is automated systems, substituting many semi-skilled jobs present throughout administrative, labour or customer relations roles in automotive, banking, health, insurance, accounting and call centre industries which are typically outsourced to developing countries such as India, Vietnam, south Africa and Morocco. Poorer nations have used their relatively low-cost labour force as a driving force to catch up with developed economies, examples being China, Thailand and Vietnam. This method of outsourcing may soon come to a close as the growing influence of AI and supplementary technologies like 3D printing will encourage consumers to manufacture locally with the convenience of customisation. The IT industry also encourages skilled workers to migrate to safe and developed cities that offers career opportunities like Silicon Valley, but exacerbates skilled labour in developing cities, further driving inequality between regions. It is critical that AI is used to serve the broader society such as boosting productivity in developing economies and enabling greater access to education, health, employment and other opportunities. MPesa and M-Tiba serves as a great example by facilitating mobile phone-based money transfer and health services in Kenya. It is crucial for policy makers to confront the rapidly and far reaching impacts of AI and make decisions that will benefit society. Economists say AI is in a different category. The rise of AI is the drop in cost of prediction. Prediction: taking information you have to generate information you don’t have. e.g: Demand forecasting, classification. The cost of prediction is plummeting, and the implication is that we will be using significantly more of it (better faster cheaper), we will start converting non prediction problems into prediction problems (autonomous driving was once seen to be impossible because it would require programming of an infinite number of variables in an uncontrolled environment. This has since changed with AI which uses predictions of a safe driver’s actions. Language translation was once rule based issue requiring expensive resources such as linguists, but is now a prediction machine. Google will soon launch a commercial translation device called google pixel buds, which uses AI). Input, Prediction, judgement( how costly would it be for the prediction error), action, Outcome, Feedback. Implications for jobs in the economy. The substitute machine intelligence (Value of human prediction falls as the capability of machine prediction increases and its cost decreases. Human prediction is very flawed, it has systematic biases and is very slow and leads to failure often and thus future businesses will be likely to incorporate machine prediction over human labour, a bleak view of the future. However the value of other factors like Input, Judgement, Action, Outcome and feedback will increase as a complement of Prediction. This means that data will become more valuable as an aspect of input, since it can be used with variety as a result of cheaper prediction. Our judgement increases dramatically as the value of machine prediction increases, a doctor can feedback data greatly improves the prediction quality and informs better judgement overall. Combining offline and online data and enables the organisation to stand apart from major players like google, facebook, amazon etc. AI as a prediction tool is actively increasing its accuracy and quality of prediction at a dynamic pace, and could have far reaching impacts that may surmount to a disruption in business strategy. This notion is best outlined in Amazon’s endeavours to transform its online shopping to shipping business model through the use of AI. By gradually improving the prediction accuracy of product recommendations to customers, Amazon could eventually find success in its patented and tested anticipatory shipping strategy which is to ship the product to the customer before they even shop for them, providing convenience to customers, upselling additional Amazon products, simplifying logistical processes and gaining the first mover advantage. This profound disruption of the conventional online shopping model through AI, serves as one potential example in many other major transformations that could occur in Recruit’s various business strategies that may significantly enhance its ribbon model. Through data, businesses could forecast certain days or specifically certain times of peak demand from customers to Recruit’s restaurant and salon clients. This information could be provided to assist store owners to allocate the appropriate resources and staff during busy periods and also highlights certain opportunities for clients to take advantage of. Cultivating an online and offline data generating environment through Recruit is the foundation to AI applicability across all current and future business activities This is because feedback data, informs greater accuracy in predictions and is often difficult and costly to collect. The success of Recruit’s sales team in enriching relationships with their clients has allowed access to deep offline data, such that the outcome of online advertising campaigns could be measured reliably through the boost in sales. This efficient input and feedback system that Recruit has developed with their clients, serves as an ideal foundation for accurate predictions which holds an advantage from major players like Google, Facebook and Amazon. This will likely accelerate Recruit to be one of the first beneficiaries in an AI induced business disruption. Like Amazon’s anticipatory shopping strategy Recruit’s client to customer matching business strategy, such as the AI applications in SalonBoard’s service can experience a major disruption as AI plays a larger role in the future. As the prediction quality increases with a growing wealth of consolidated input and feedback data, Recruit can begin to accurately predict when a user will require a haircut and even the specific hair style, weeks or even months ahead of the user’s inception to get a haircut. Subsequently all appointments could be automated on behalf of the user, which completely expedites the booking and scheduling process away from the client, ultimately improving scheduling efficiency, resource and workload management and allowing the clients to “do what they do best – look after customers”. From the user’s side, the inconvenient cost of searching for the right salon as well as booking an optimal time is mitigated through prediction, further improving the relationship with the Salon as well as gaining the reduced-price advantage derived from the Salon’s cost savings as a result of operational efficiency. This shared reduction in cost between the user and client driven by prediction, will entice more client and customer participants as well as establish Recruit’s position as a “reliable friend for the end customer” and simultaneously increase dependency from their clients. This transformation is effectively applicable to all business segments across Recruit, especially in the profound enhancements in HR services, as the process of marketing to a large range and narrowing down to the best candidate, is ineffective with biases, and costly overall. This convention can soon transform into an approach where candidates will have been selected through prediction algorithms. Since Recruit’s acquisition of Indeed, the organisation has aims to improve job matching through the adoption of AI, using input data from candidates and employer’s proximity, salary estimates, ratings and reviews, and so on. In its strategy, Indeed employs natural language processing to the listings on its site, optimizing unstructured data into text that appeals to the right candidates and has since increased the rate of employers responding back to applicants by 10 percent across all industries. Furthermore, the organisation aims to enhance the ribbon model of matching employers and candidates through AI by evaluating candidates’ résumés and predict their suitability to the role requirements and provide candidate evaluation models for employers to issue to candidates, to further improve optimisation and reduce cost of examination. Essentially the emphasis in AI intervention for the HR platform is reduced cost, as employers will gain access to a larger and more desirable pool of candidates that can reach towards specific talent across country borders, as well as anticipating promising individuals undertaking education and training which before AI, restrained this possibility. Grasping this new technology into their HR segment can push Recruit into the forefront of the staffing industry and fulfilling its ambitious goals of reaching the largest number of positions filled. Other creative ways to incorporate AI to improve Recruit’s verticals would be open source code like image recognition which could be significantly helpful for fulfilling a customer’s specific requests by uploading a photo of a specific hair style or food type to then be recommend with an appropriate client that can fulfil the specific request. Upon discovery of AI’s application in HR and combining with Salon board, we can really see the profound far reaching impacts of AI to the overall ribbon model. Indeed’s candidate evaluation prediction service will always be constrained to an assistive role and must be subject to greater scrutiny. This is because the algorithm may not perform accurately in the real world and could commit unintended unethical acts to perform its duty. Partaking in AI into business strategy must align with corporate ethics and values and broader societal morals, which suggests that AI must have extensive training with humans and is likely to play an augmented role over a primary one. Many examples of AI’s misuse have emerged recently such as its involvement in recommending vaccine misinformation, in the form of articles and youtube videos to vaccine sceptical viewers, which progressively compelled many to become adamant antivaxers today. The moral externalities that is not factored in to Ai which lacks inquisitiveness, is evident here as it’s aim to keep viewers on the platform and increase ad viewership is also noticeable in a growing trend where political opinion pieces are attached with more extreme, inaccurate and misguided material such as crackpot conspiracy theories which has been proven to attract viewership. 22% of net sales are sourced from Recruit’s targeted advertising and marketing revenue from their platforms. With a significant portion of the revenue share coming from targeted ads, Recruit must be capable to factor in ethical controls where necessary to avoid any forms of discrimination or spread of misinformation. This was exhibited where higher salary job postings on google, were disparagingly less visible for females and google search only reflected 11% of the actual 27% of female CEOS in the industry. It is strongly recommended for Recruit to view this case as a central concern in using prediction models to recommend clients with customers and mitigate any scenario where discrimination may occur. As recruit’s operations begins to be more involved with AI and it’s capacity to replace rudimentary roles, it is crucial to be aware of its potential in driving inequality across other regions, and displacement of jobs. Recruit’s transition from a paper magazine medium to the internet has in part reduced the demand for printing services but also increased the need for jobs in cybersecurity, webdesign etc. AI can allow recruit to access developing country markets with a superior unmatched service and uplift many semi skilled labour in the process. The IT industry also encourages skilled workers to migrate to safe and developed cities that offers career opportunities like Silicon Valley, but exacerbates skilled labour in developing cities, further driving inequality between regions. Boston consulting and MIT published a report that Ai is creating both fear and hope for the general workforce. The industrial revolution and the computer revolution has both produced significant transformation in jobs which has also increased in variety. d Interpretability Data citizenship Fairness Governance Future of work Most users in Recruit’s matching platform implements This will unfortunately drive inequality if HR boosts the connection of workers to developed economies etc etc Prediction machines utilize three types of data:(1)training data for training the AI, (2) input data for predicting, and (3) feedback data for improving the prediction accuracy. Data collection is costly; it’s an investment. The cost of data collection depends on how much data you need and how intrusive the collection process is. It is critical to balance the cost of data acquisition with the benefit of enhanced prediction accuracy. Determining the best approach requires estimating the ROI of each type of data: how much will it cost to acquire, and how valuable will the associated increase in prediction accuracy be? Statistical and economic reasons shape whether having more data generates more value. From a statistical perspective, data has diminishing returns. Each additional unit of data improves your prediction less than the prior data; the tenth observation improves prediction by more than the one thousandth. In terms of economics, the relationship is ambiguous. Adding more data to a large existing stock of data may be greater than adding it to a small stock—for example, if the additional data allows the performance of the prediction machine to cross a threshold from unusable to usable, or from below a regulatory performance threshold to above, or from worse than a competitor to better. Thus, organizations need to understand the relationship between adding more data, enhancing prediction accuracy, and increasing value creation. Unknown Knowns are a common underlying issue affecting prediction accuracy Data can inform decisions as well as come from decisions. If machines do not understand the decision process that generated the data, its predictions can fail. This can arise from reverse causality The process is an excellent predictor that can enhance 1. Good predictions (data that is used to make the decision) 2. Reverse causality (data that was derived to trace back from the decision that was already made) 3. Omitted variables (data that is not yet known that contributed to the prediction) Reverse Causality: The feedback data on an action not taken, can often be unattainable as many circumstances only allows for data to be observed ensuing the action taken. Disadvantages of AI