EU standard for the Full Stack of an Artificial Intelligence ( AI )

The proposition to start a full stack Artificial Intelligence standardization as an European Union Project.

 

To:                        EU AI Alliance HLEG

From:                 Kai Salmela, on behalf of ROBOCOAST R&D Center of Excellence, Pori Finland.

Date:                  19th October 2018 

 

 

Background:

Today as artificial intelligence development is leaping fast forward, there is a great need for guidance and standards that we could follow in order to get the most of this progress.

Europe has not been the leader in the development of an artificial intelligence, that is why the biggest AI companies come from the USA and the most advanced AI products come from China today.

This is not the best possible situation for an AI, as Europe is mostly absent of the development and some crucial points are not taken care with one of the best European tradition – Standardisation.  International standardisation has just begun, but it seem, that it concerns only parts of an AI – not all of it.

Now that we have an EU AI Allied forum for discussion and propositions, ROBOCOAST would like to make an input. 

 

Proposal:

European Union should take a lead of an AI standardization and see that every aspect of it is defined with the satisfaction. 

This mean Full Stack Standardization of an AI, where all of the technical-, ethical-, definition-, security-, educational-, user- and future- aspects and concerns will be thought thru and defined.

This is not an easy task. It will need lot of know-how of an AI, but also of an EU Legislation, EU Standards, International Standards but also Ethical, security and medical expertise. There really is not another single operator than European Union who could succes with this task with the global community.

There will be unseen development and all new problems to solve, but the right conduct will help us all to get better, more affordable and safer AI in the future.  We all know that AI standards will come anyway, so why would we not do it in Europe? After all, this is the home of standards.

 

 

Benefits:

 

A Definition.

Today AI is very widely used buzz-word that can mean just anything. AI can be defined in a standard, so that we all know what we’re talking about and a definition ‘what is an AI’ would help also EU when funding is proposed to any AI project.

 

Faster Development.

Everyone can see how much quicker development has been for car- or cellular phone- industry once standards were placed, and this is now that we need to do  for an AI industry. The pace of development needs to get on a higher level, and it is not happening by the free competition without any guidelines.  Everyone will benefit on this on a long run.

 

Affordable AI.

AI will be more affordable and can be reached by more of people than today.  Once there’s standards for each of the part, there will be more companies offering compatible hardware, software, algorithms, education and more of people will be able to create intelligent systems and even AI products to different needs.

 

Safer AI.

When standards will be set, security, control and traceability are among the most important definitions that it will state. European GDPR and equivalent laws around globe need to be oblicated in the operations of an AI. People need to know that their information, life, money and friend is in safe hands with an AI. Furthermore, if it makes a mistake, the reason can be traced down in a standard way.  This part of standard is really important, and it needs to be defined before we start to use AI to develop better AI. – after that, tracing and correcting the process can prove to be impossible task.

 

Better readiness for the Future.

After a proper standardization, AI will be better ready for the future. There will be new innovations, perhaps quantum-processors that change the game, and still there would be guidelines how to develop AI with them so that it is compatible with the older versions.

 

Democracy.

Doing this work within EU make this process more democratic, when anyone can comment and state their concerns of the latest iteration.  It means also that AI standard will not be cut into stone by any company and there will be a process to enhance and change standard when needed.

 

 

 

 

Some cons:

Resistance.

There is a change, that there will be companies who fear to lose their power as the AI will develop. So there will be an opposition to standardization but once the benefits are clear for the greater public, opposition will fade away eventually.

It is a big task.

This is a huge task and it will take a lot of time and effort to succeed. With this EU has a need to a global co-operation in order to bring standard AI available for everyone.

The race of AI standardization is already on, but only for limited field of it.

Full Stack AI has not yet been recognised as a need, and soon it can be too late to start this work if other standards are not compatible. Standardization need to take a step further and see the whole picture of an AI , since this is much more than equipment or tool – it will define our future.

 

For the end:

Robocoast has a great hope that this standardization can be started now.  We have not all the answers for everyone, but we hope to co-operate with you in order to solve all the questions. We are ready to start this process, and willing to propose co-operation also to the Government of China if the agreement is found in EU.

 

Kai Salmela , AI Specialist in ROBOCOAST R&D Center

kai.salmela@prizz.fi        

P.O Box 18, Siltapuistonkatu 14   FI-28101  Pori,  Finland

 

 

ROBOCOAST is a Center of Excellence in Finland governed by Prizztech Ltd. 

Robocoast R&D Center - Robotics Living Lab for Companies -project is funded by the Regional Council of Satakunta (ERDF) as well as Pori and Rauma regions.

   

 

 

        

Tags
Artificial Intelligence AI standard future AI global importance

Comments

User
Mibgħut minn Peter Brown f’din id-data:Mon, 12/11/2018 - 13:24

My main question is: why?

There is already an initiative to develop a "full stack" of AI Standards, through the ISO system at a global level (ISO/IEC JTC1/SC42). Why don't European organizations get engaged here at an early stage and help shape future standards globally rather than running off into a corner as they so often do, and with limited results?

 

In reply to by Peter Brown

Profile picture for user n002daga
Mibgħut minn Kai Salmela f’din id-data:Mon, 12/11/2018 - 13:48

Thank you for your comment, i do appreciate different opinions .

Main reason is the historical lack of understanding of the Global needs within ISO.  If only all of the countries would partizipate with their full knowledge and power into ISO standard Development , it may be possible, but they do not.  I've also learned that members of ISO project have been exceptionally opinionated agains subjects that have been raised within EU AI project.  My company is in at the ISO standard project, from our national standardization association and we do our best - ofcourse.   

Biggest AI houses are from the USA , and they have a lot to loose in this standardization. Smaller cultures and their needs just do not matter so much. Within EU we still have a better possibility to gain unbiased , culturally diverse , modular , safer , more affordable AI with the proper standardization. And there's also a guestion of choise:  If you need to choose AI for your company, it will be either the leader of the industry or by right standard. Which one would you choose if you're european factory owner?  A Full Stack AI standard is'nt just a technical execution, it really concerns all around AI and it needs to be treated as all inclusive set of guidelines and example system - a standard. ( today the standard has been shrunk to tech only )

Unfortunately EU is now developing AI guidelines to proglamations and laws, not for standard. Standard would be much more influential globally. That we are about to change.  You're welcome to support this effort.

 

wbr Kai Salmela