Feedback to Peter Rudin's essay:   HUMAN-CENTERED AI TO ACCELERATE EDUCATIONAL INNOVATION

excellent essay!

Unesco Planning Education in the AI Era: Lead the leap Beijing (May 2019) referred to  in the essay. Interesting Unesco approach, still very high level, does list positioning criterias, set some initial frame and recommendations. Good to see global institutions do also engage, anticipate the AI impact.. 

https://en.unesco.org/news/first-ever-consensus-artificial-intelligence-and-education-published-unesco  - report language versions - ( SDG4)

MOOC & knowledge access actually at unprecedented scale assisted by AI but also student tier collaboration (all people with access to internet). Actual standards are defined by giant companies with their preferred  AI stack (= strategy ) and open source initiatives. Government backed standards are lagging behind (new metrics 'AI lightyear' are not yet defined for measures, certification and government regulations. Currently a self-regulated process, very similar to the one for software development since many years (Programming Language, framework stacks linked to giant companies but also strongly supported by the developer-user community. This has various reasons, some prime reasons: in a fast changing context only the hands-on people can provide ultimate support, but moreover these people want to protect their skill asset obtained via hard work and trainings. Also targeting maximal sustainability of their work and acquired skills (strongly linked to used methods and frameworks (minimize too frequent refactoring, legacy replacement of complex systems and related costs)) .

Mary Meeker internet report 2019 has additional information on education and MOOC (education page 234), AI (page 81, 147, 201). Her report is a yearly key reference (metric) for me too.

(Great MOOC personal experience, best so far 2013 Start-up engineering (Stanford),  Meteor 2015 (London University / via Coursera). 

Some USA attempt to establish AI standards (kind of prerequisite for education too)  NIST.GOV    

The +6h video recording of the sessions outlines well the challenge between Government standard bodies and private giant companies. (By definition it is complex, needs good balance and tolerance: standard definition, certification & education versus fast paced private companies innovation which can't expose all their R&D results but need to secure investments and keep leading).

Draft plan (initial outcome): Recent memo       Report   

The request for AI standard was requested by the US government probably also to defend US against China (the party who defines and maintains standards can impose more control). 

IT development - Open Source

Robotics, automation (AI). Code audit used during MOOC or during the hiring process (fully automated coding tests, results) is used established since years.

Open source large scale stacks (personal available time is the only limit for people with internet access who want to self-train and develop themselves,  prototypes, fast publish for fast failure or success. Education programs need to include these methods and movements. The efficiently combined forces will become more important to solve human existence related challenges in future. 

Gnuradio (e.g. security pentesting or IoT virtual device design-configuration prior assembly. Large amount of shared components to re-use and contribute)   

Knime (e.g. data science, AI, BI etc) large number of nodes for reuse and contribute)

Meteor full stack (it takes 1/2 hour to create a scalable working prototype via the Meteor full stack. (Going visual, with meteorkitchen the coding complexity is also removed. /  active community)..

AI Giant companies continue the immense effort to establish their own AI stack as industry standard. 'Open Sourced' to bind developers, users. Alphabet tries with succes, since years, to repeat with tensorflow (AI) what they did with Android. Then bind, blend their assets (e.g. native tensorflow AI on mobile for exponential impact, guaranteed scaling (getting more and more developer and consumer locked-in).

It is a (temporary!) win-win situation, software and device manufacturer need some minimal sustainability and exposure (scale) of standards which can only be established by giant-companies but need increased governance as you discussed and analysed in various essays (governments, independent bodies).

US universities are since many years much interwoven with the industry (students when leaving the university are very strongly conditioned, aligned to company specific needs) 

Tensorflow playground description

Tensorflow playground 

CrashCourse 

Other recent samples of open-source business control-convergence: IBM acquisition of redhat, Microsoft acquisition of github 

 best greetings and many thanks for the excellent essays

Hannes