House of Lords All Party Parliamentary Group the Enterprise Adoption Of AI 21 Oct 2019

Panel Representation from REC, Rialto, KPMG, PWC, Goldsmiths Uni

Observations and notes from the meeting:

Exec Summary

The business case is missing the negative impact that we can’t manage: too many issues raised with weak and conflicting analysis. Dissent even at Gov level is squashed e.g. “ we can’t cover all of the issues!). The Big 4 dominate these APPG panel discussions and the tech discussion. The big4 are conflicted, their analysis is weak conflicted and makes little sense:

·      pressure to push the tech gravy train

·      regardless even when it does not look like tech is good for their client’s business nor society.

Macro picture

·      What does good look like? Who is accountable?

·      People are losing our ability to stay in control, exponential development rate and fear of missing out, and becoming addicted / excessive time to the connected society

·      UK ONS 1m fewer jobs paying more than £100k and the skills required are greater

·      Big 4 analysis is weak, what is their motivation to sell / not? No macro picture: conflicting analysis / No view of big picture or indication of viability. Overselling upside?

·      We still know so little about people and people well-being, tech plays to FOMO fears 42% have no tech business plan

·      Impact on reducing wages, due to relative reduced human value add

·      Focus on product commodification to enable digitise reduces value add

·      Reckless statements; KPMG £630bn to pay for £90bn retraining costs

o  Who will pay for mass retraining?

o  Won’t be able to train fast enough anyway – even the techies

o  There are near term techie job losses due to the next wave of chip design / 5G

·      PWCs 2018 report needs challenging underlying assumptions look flawed

Competitive picture

·      Customer focus – bots, and intensity of surveillance, often no people contact option

·      Parasitic vs value-adding? 90% of Fin Tech’s are payment orientated

·      Revenue reductions – value and profit loss

·      Loss of competitive advantage- digitisation makes it easier to switch

·      Loss-making startups “destruction vs disruption”, anticompetitive and free of charge services – giving away sustainable sources of value. Paradoxically anticompetitive legislation was based up overcharging not giving value away!

Society impacts without solutions

·      Odd utopia positioning: middle management: “define dream job”, how to achieve using tech, open possibility, growth mindset (abused and normalised!)

·      Operating model – move from silo – fluid, customer-facing

·      Flawed unaffordable assumptions UBI “Leisure” – no one will pay for people to do nothing reducing tax collections will not make this possible

·      Can’t reskills those with lower IQs: people will become marginalised from work

·      People first vs cost-benefit

·      Government debate closing discussions around managing the downside

·      Industries such as cars will employ fewer workers – simpler products electrification vs combustion engine

·      Do we all want to be techies?

·      Augmentation vs automation of roles – the interface is damaging manipulation, polarising, generalising;

·      Tech in recruitment – will create generalisations from data, people are not standard? Trends to commoditisation of labour

·      Increasing labour dissatisfaction with roles 35% fed up, only 9% of young want a promotion!

Business Risk

o  Sophisticated AI – Cyber-attacks – won’t be able to cope

o  Google vs amazon moist advanced AI, collaborating with Chinas JD.com profitless growth – focussed on zero jobs!

Regulatory

o  How can we protect ourselves from China? Without having to follow their race – bottom

o  Ultimately competition of tech intermediators is a race to the bottom: price fight

o  ICO / FCA Regulation will lag – and never catch up, smoking took 50 years! Needs a new regulator

o  GDPR, is unrealistic meaningful standards cant cope with machine learning – the reason why only recourse refer to a person, people won’t understand the decisions /output

o  governance explain-ability how can tech decisions be understood and rationalised?

Operating issues

REC Tom Hadley

·      Job role most important decision someone makes

·      Jobs slow being removed from the market

·      The focus is on specific roles and tech reinforces

Questions

·      Who will manage the relationship

·      Interviews robots – love it!

·      Sifting

·      Regulatory aspects – H&S etc, gov regs

·      Equitable fair bias

Maria Axente PWC

Slow adoption only 6% large scale adoption, 5 key factors:

1.      What is the tech org structure, innovation team

2.      People upskilling

3.      Trust – autonomous bias, global cultures are different, who is responsible?

4.      Data strategy, leveraging dormant data, open new opportunities, not just problem solving, create value?

Role of leadership; middle management is key

1.      Art of possible

2.      Shared vision

3.      Risks / security / economic

4.      Social risks

5.      Augment jobs, work together

6.      Ethics

IT is left out of the discussion – really?

This is a link to my challenge to the PWC AI impact summary of 2018

Chris Brewer Goldsmiths Uni

Microsoft 6m research

1.      48% are experimenting

2.      8% scaling

3.      31% not using AI at all

4.      Use grew from 11-24% last year

3 areas

1.      Experiment and democratisation

2.      11.5% better results (measures?)

3.      What should AI do vs can do?

4.      ROI

5.      Reskilling

Adult learning will be an ongoing requirement plan 10 years ahead

Tech makes bias visible vs reinforces bias, and focus to create steps to resolve it

Richard Chiumento CEO Rialto Rialto APPG evidence summary for 21.10.2019

1.      The pace and progressing exponential rate

2.      270% increase in implementing AI

3.      Final innovation – end of the human race

4.      50% of fortune 500 companies have gone, lack of digital – or digital?

Warehouse employment comparison: picking 27000 orders; now employ 40 vs 600 staff originally

Last year 16.5bn investment in AI

There are now 1m jobs that pay more than £100k

Conflicting picture in terms of work impact, WEF, Oxford Uni, PWC, & KPMG

IBM estimate 120m will have to be reskilled

“Leisure” – who will pick up the tab?

Stagnating wages,

1.      1 robot per 1000 workers reduces wages 5%

2.      Average skills vs skills greater gap than ever

Recommendations

1.      C suite director reskilling

2.      Lifelong learning, importance to keep up

3.      Inconsistent in skills learning must be faster

4.      Contradictory evidence

5.      Tax robots

6.      Legal obligations to reskill – who will pay for will cost 90Bn per 1m workers

7.      Each individual take own responsibility, stay curious keep up / give up!

8.      MIT suggests plan personal development ahead 10 years

KPMG Justin Anderson

1.      Value impact of data and algorithms to company valuations

2.      £630Bn tech upside to UK economy??! – another conflict

3.      Middle management is key, often left out, Kanban systems etc help

4.      Accelerate

5.      Define successful outcome, productivity/efficiency

How to push

1.      Data systems

2.      Outside in

3.      Inside out

UK framework

1.      Made up of Ocado, gov officials, academic

2.      Need multi-industry input

3.      Women?

4.      agility