Dear AI alliance members,

We hold our second meeting last week in Brussels and made progress on the structure and content of our dedicated contribution on skills and talents. Please have a look at this (very) raw draft and feel free to comment and contribute in your comments. Deadline: 27 November. 

If you have some elements for the recommendations (we'll finalise them during our next meeting in december), please make sure they are as concrete and quantifiable as possible.  

Thanks to those of you who have already commented my last post. This was very usefull for our discussions. 

Up to you now!

Draft structure of the contribution

• Introduction and Objective
• Target groups (5 distinct target groups with specific challenges)
• Cross-cutting dimensions
• Recommendations for 5 Target Groups (long-, medium, short-term & addressing different stakeholders at EU-level, member-state level)

Why are Talent/Skills and Education important?

• Talent/Skills is Key Enabler for AI Ecosystems (industry) but also for world-class research (keyword: talent management and retention)
• Actualization is one of 6 core Values for Trusted AI: part of this value is life-long learning and education
• Major transitions in labor markets: reskilling and upskilling necessary to prevent unemployment;
• Future of work is a key area to invest (collaborative robotic, assistive systems, augmented work environments, etc.)
• The education component in the Horizon Mission oriented innovation policy

What needs to be considered?

• Talent retention
• Equality, Diversity & Inclusion
• Hybrid qualifications especially for inter-disciplinary research (which may require changing rules/legislation)
• Flexible curricula for deep specialists and broader generalists; also inter/trans-disciplinary
• Combine research/education and working in industry (gear towards needs of industry vs. engineers need to be trained for various industries) Complementarity: companies must invest in their human capital; incentives need to be provided
• Timeframe of university curricula is different than what industry needs
• Measureable targets (qualitative & quantitative) & specific Recommendations

Objectives:

To create an AI skilled workforce, to address the societal changes AI will bring and ensure that AI development and deployment in Europe is not restricted by a skills shortage, to ensure that citizens are educated about the impact and consequences of AI and empowered to benefit therefrom.

• Complementary human machine approach; AI Systems should augment capabilities of humans
• Retention of talent in Europe (prevent brain drain, attract talent): attractive science careers with higher salaries and better research and working conditions, access to funding
• The upskilling should concern everyone, from initial education to continuous VET, from children to adults, from citizens to workers and employers. .

The upskilling should not only be focused on university degree but also other type of degree for low qualified workforce​

Target groups

We have identified 5 different target groups; Each target group requires different/specific skills-sets and is facing specific challenges. Thus we need specific, tailor-made recommendations and education programs for each target group

1. AI Specialists (provide infrastructure/data for and develop/implement AI Systems)

• Developers (design, programming, etc.)
• Entrepreneurs (implementation: management and leadership skills)
• Data-Managers (data curating, analysts, etc.)

Skills for people who develop & implement AI Systems

Challenges identified:

• more AI deep talent but also more other fields’ experts confronted to AI
• Need to develop and retain more European basic research and researchers on AI. Laying a solid ground for this by better teaching mathematics.
  • hybrid qualification: inter- and cross-disciplinary trainings, permeability of disciplinary silos
  • generalist vs. specialists – ambidextrous qualification, integrating both breadth and depth in skills at the same time
  • skilled in ethically aligned design/responsible design
  • People need to be educated and trained in skills in which humans outperform AI Systems (social skills, creativity); not AI related but complementary skills!

2. AI Workers (people who work with AI Systems)
3. Workers
4. Apprentices

Skills for people who work with AI Systems

• Gathering domain specific skills
• Implementation of lifelong learning
• Vocational training and upskilling

Challenges identified:

• Unlearn to learn (impact of assistive AI systems on cognitive skills unclear)
• Paradoxes of automation (assistive AI systems may lead to deterioration of worker skills, out-of-the loop scenarios, etc.). Examples were given by AI alliance contributors that US Navy recently re-started to train its officers to navigate by the stars since they fear relying on GPS navigation makes ships vulnerable to being useless when confronted with a cyber attack. Also, planes could fly on their own already for many years but there are still two pilots on board and, following recent incidents, the FAA in the US even required airline pilots to fly manually more often to maintain skill levels.
• Speed of innovation; up-to-date knowledge/skills, also for the teachers in vocational training who need frequent train-the-trainer measures to ensure their up-to-date training capability.
• Awareness raising for training policies stakeholders, i.e social partners at all level and policy makers.  
• Defining training needs: Depending on the company and its (potential) field of AI application, companies need different skills and thus specific/individualized training sessions.
• People need to be educated and trained in skills in which humans outperform AI Systems (social skills, creativity); not AI related but complementary skills!
•  

3. Citizens (consumers)

Skills for people who consume AI Systems

1. Digital literacy: understanding the impact of AI System use, responsible use
2. Freedom of choice
3. Access to AI basic knowledge, training and education for every citizen on human-machine collaboration.

Challenges identified:

• Requires transparency of AI systems and information about AI Systems
•  freedom of choice vs. regulation to protect consumer interests
• Role of media (providing information, science fact vs. science fiction, fake news…)

4. Vulnerable Stakeholders
   1. E.g. children, people with (mental) conditions, etc.
   2. Support for workers who lose their jobs and cannot/do not want to be up- or reskilled

Skills for people who are otherwise affected by or become vulnerable through AI Systems

Challenges identified:

• Ensure that vulnerable groups benefit from AI systems that workers whose jobs are affected by technological change get support (career guidance, social protection, training, etc.) ; ethical issues

5. Policy Makers
   1. Government & agencies
   2. Social partners
   3. Stakeholder and interest groups
   4. Juridical system (judges, etc.)

Skills for Policy makers & social partners

Challenges identified:

• Complexity of AI implications (general purpose technology)
• Develop of “culture of data” in the society (public services, companies, civil society) in order to allow AI to help resolving economic, social and environmental challenges while respecting privacy and human rights.
• Speed of innovation
• Shifting participation
• Possible specific recommendations made by the HLEG to European social partners when they will start negotiating on digitalisation (from 2019, tbc).
• Inertia in public institutions: tend to focus on old/traditional sectors and not on new areas: recommendation is to form digital committees in parliament

Cross cutting concepts

• Gender & Equality (Diversity & Inclusion): attract female talent to STEM (today only 20 % in STEM are female); discussion not only focused on gender but also including other minorities, immigrants, etc.
  • Understanding technology as inherently gendered and its impact on design
  • More women in STEM fields may (partly) resolve shortage in talent in these areas
  • Typical female occupations (administrative jobs, merchandising, etc.) have high automation risk; vulnerable group
• People with disabilities
  • tbd
• Cultural transition:
  • What is work; the meaning of work is changing; evaluation of work
  • Schooling systems today are “disciplining machines” which prepare people for industrialized work, subordination into hierarchies and standardized processes; skills needed in the future are analytical thinking, creativity, social skills, etc.
  • Cultural transitions are difficult and may take very long
• Time frame (short, medium and longtime perspective):
  • Changes affect whole educational system (from pre-school education up to tertiary, post-graduate education, and continuous VET)
  • Adaptation in educational systems in member states have to be initiated now to be effective in 10-15 years;
  • Short- and medium term strategies have to be developed to deal with pressing issues now (transitions in labour markets, re- and upskilling)

Recommendations & Stakeholders

Overall Timeframe for Recommendations: 2030;

Short-term, medium-term and long-term recommendations

Stakeholders: EU, member states & other Stakeholder groups