Artificial Intelligence
How does Watson work
Uses human learning process of observing, interpreting and analyzing in order to "achieve mastery" and "develop expertise"
What is the real issue with AI and jobs
Very few entire jobs will be automated Real issue is the percentage of work tasks that will in the future be done by AI and robots Such tasks are likely to be spread across most manufacturing and service occupations at a wide range of levels and covering many professions e.g 20% of task automation may impact 95% of workers and result in 10% of jobs being lost
Watson Products
Watson Virtual Ages (used by RBS) Watson Analytics
Watson Virtual Agent
Watson integrates with other applicatoins to help customers with new questions Watson delivered through cloud so always on current version always getting better without user having to do anything helps implement effective customer self service improves first contact resolution and reduces costs
Convergance
AI increases the potential to replace labour with technology (Capital)
Digital assistants
Alexa, Cortana, Siri and Goodle Assistant developing into the next customer interface
Cloud AI services
Are democratising AI by allowing it to be used in all manner of products and services by relatively non technical people
What can AI be in next 5-10 years
As important an interface as the web and with major implications for marketing
AI Customer Service (IBM blog, 2017)
Can deliver a level of responsiveness that isnt humanly possible only has to be trained once always on. Reliable. Can be personalized Millenials prefer self-service that doesnt involve a human being or a phone by 202 85% of customer interactions will be handled without a human agent
Three approaches to AI
Codified rules (e.g expert systems Heuristics and evolutionary learning Artificial neural networks (ANNs) and deep learning
New forms of employment
Historically, job losses due to tech have been offset by creation of new forms of employment and ahve raised living standards but there are no guarantees that this will happen again and new job creation may lag old job destruction there is risk of premature deindustrialisation in many nations
Watson Group
IBM created it in 2013 as a $1bn business unit "dedivated to the development and commercialization of cloud-delivered cognitive innovation"
Key western market players
IBM-with watson Microsoft with Azure Cognitive Services Google Cloud AI on the google cloud platform Amazon with Machine Learning on AWS
AI denial
Many business gurus claim that AI doesnt exists and will only impact a narrow range of jobs or will mainly augment Only repetitive tasks will be automated so freeing workers to do other things also assumes that most people can and will be able to transition to higher more rewarding activities in contrast tech gurus fear mass unemployment
Jobs at risk?
March 2017 PWC suggested that 30% of UK jobs are susceptible to automation by 2030 Sep 2017- RSA/YouGov UK survey reported that on avg, business leaders believe 15% of jobs may be automated Using World Bank data in 2016, Citi and Oxford Martin School stated that 57% OECD jobs are at risk of automation
AI benefits and applications
Resource savings via improved logistics removal of language barriers improved medical diagnosis facial payment systems, production quality control systems, tracking systems that utilize AI/ML vision recognition
Davenport & Kirby (2015)
Suggest that humans will remain employed if they step up step aside step in step narrowly step forward Argue that AI will mainly be used to augment rather than to automate so deepening human work, rather than diminishing it h/e their case examples illustrate more-productive human AI teams with fewer people employed
New Kinds of Jobs
Wilson et al 2017 report an Accenture study of companies using or testing AI and Ml they report three new categories of job Trainers: e.g. customer language tone and meaning trainer Explainers- e.g transparency analyst Sustainers- e.g "automation ethicist" and machine relations manager
The work most at risk
Work that can be codified into standard steps and which is based on cleanly formatted data this includes compliance and tax as both are rule based and data intensive but all tasks based on digital exchange are a candidate for AI automation Some people with best digital stills may hence in the future be least empoyable
What is AI
computing technology with cognitive capacity that may substitute for some HI to date computers have made arithmetic fast, cheap and abundant, AI machine learning will do the same for prediction this will increasingly facilitate speech interfaces, vision recognition systems and the operation of autonomous vehicles and robots