EU to Constrain Certain AI Uses?

To be expected reaction from EU regulators.

Facial Recognition, Other 'Risky' AI Set for Constraints in EU

Bloomberg, Natalia Drozdiak, April 21, 2021

The European Commission has proposed new rules constraining the use of facial recognition and other artificial intelligence applications, and threatening fines for companies that fail to comply. The rules would apply to companies that, among other things, exploit vulnerable groups, deploy subliminal techniques, or score people’s social behavior. The use of real-time remote biometric identification systems by law enforcement also would be prohibited unless used specifically to prevent a terror attack, find missing children, or for other public security emergencies. Other high-risk applications, including for self-driving cars and in employment or asylum decisions, would have to undergo checks of their systems before deployment. The proposed rules need to be approved by the European Parliament and by individual member-states before they could become law.   ... " 

The Value of Constraints for Creativity

 Even if you don't embed them formally, we found, as in an optimization.  It states the walls to bounce off of.   Note the use of 'some' in the statement, which then requests the next word:  which?

The Role of Constraints in Creative Problem-Solving

Paper:  https://www.hbs.edu/faculty/Publication%20Files/21-068_2425a8e8-7eaf-47fa-8db3-4061c303f606.pdf  

by Daniel Ehls, Karim R. Lakhani, and Jacqueline N. Lane,  in HBS Working Knowledge

This study shows that constraints can support creative problem solving in a consumer electronics setting. Adding (some) constraints increased the quantity and quality of strong ideas generated and selected through an open innovation process.

Author Abstract

The role of constraints in the problem solving process has been a central line of inquiry in the creativity and innovation literature with ongoing debates of whether constraints imposed on creative problem solvers diminish or enhance their efforts and outputs. We investigate this question by designing and executing a field experiment in collaboration with a world leading company in consumer electronics seeking creative solutions through a community crowdsourcing program to improve the wearing comfort of their popular headphones. We mobilized 1,833 problem solvers, 331 ideas and 435 community evaluators to rate the quality of the solutions, for a total of 2,473 evaluator-solution pairs. To make experimental comparisons, we exogenously varied the number of constraints faced by the community problem solvers to determine how exposures to constraints affected the number and quality of solutions. We find causal evidence that moderate levels of constraints increase both solution quantity and quality. Compared to problems framed with no constraints, having some constraints causally increases a solvers’ likelihood of proposing a solution by 6% or 1.5 times. Turning to solution quality, we find that while constraints decrease the average novelty of solutions, they have no effect on the most novel and useful solutions. Lastly, we observe an inverse curvilinear relationship between the number of constraints and the most creative solutions, where problems with some constraints increase the likelihood of coming up with one of the most creative solutions by 3–4% compared to problems with no constraints. We discuss the implications of our findings to the creativity and problem-solving literatures. .... 

Reinforcement Learning for Skill Discovery

Can skills be dsicovered.  That is, a means to find better behavior that leads to prescribed real-world goals?   Here in the Google Research blog,  addressing the use of unsupervised reinforcement learning (RL). Note the determination and inclusion of constraints.   Like in classic optimization problems. Largely technical, but thoughtful positioning.  Considerable links in the article below.

DADS: Unsupervised Reinforcement Learning for Skill Discovery
Friday, May 29, 2020
Posted by Archit Sharma, AI Resident, Google Research

Recent research has demonstrated that supervised reinforcement learning (RL) is capable of going beyond simulation scenarios to synthesize complex behaviors in the real world, such as grasping arbitrary objects or learning agile locomotion. However, the limitations of teaching an agent to perform complex behaviors using well-designed task-specific reward functions are also becoming apparent. Designing reward functions can require significant engineering effort, which becomes untenable for a large number of tasks. For many practical scenarios, designing a reward function can be complicated, for example, requiring additional instrumentation for the environment (e.g., sensors to detect the orientation of doors) or manual-labelling of “goal” states. Considering that the ability to generate complex behaviors is limited by this form of reward-engineering, unsupervised learning presents itself as an interesting direction for RL.

In supervised RL, the extrinsic reward function from the environment guides the agent towards the desired behaviors, reinforcing the actions which bring the desired changes in the environment. With unsupervised RL, the agent uses an intrinsic reward function (such as curiosity to try different things in the environment) to generate its own training signals to acquire a broad set of task-agnostic behaviors. The intrinsic reward functions can bypass the problems of the engineering extrinsic reward functions, while being generic and broadly applicable to several agents and problems without any additional design. While much research has recently focused on different approaches to unsupervised reinforcement learning, it is still a severely under-constrained problem — without the guidance of rewards from the environment, it can be hard to learn behaviors which will be useful. Are there meaningful properties of the agent-environment interaction that can help discover better behaviors (“skills”) for the agents?  ... "   ... ' 

Theory of Agile Constraints

Newly discovered in InfoQ:

Evan Leybourn of IBM on the Theory of Agile Constraints ... 
Business Agility in general   ..... 

He defines (full interview at the link): 

I wrote about this in my article on, and with apologies to Eliyahu Goldratt, “Evan’s Theory of Agile Constraints”.
"An organisation can only be as agile as it's least agile division!"   ... 


Very basically, the Theory of Constraints is that there is a constraining factor in any process. More importantly, that there will always be a constraining factor. The Theory of Agile Constraints is that, in an organisation, there will always be a constraint to business agility. 20 years ago, that was IT. That was your software team. And that’s why it was logical for Agile, capital “A” Agile, to emerge in that domain. Today the constraint to agility isn’t IT, but rather it’s the PMO, HR, finance, or legal department.   .... " 

Abundance Theory Applied to Analytics

Abundance Theory: Had never heard it called this, and was always as a general mindset rather than a working strategy.   Analytics usually works with constraints, can algorithms take us beyond this? Can we 'feature engineer' ourselves beyond constraints?

" ..... Scarcity theory, a term coined by Stephen Covey, suggests that everything in life has its limit. Whether that thing is a spot on the team roster, a scholarship, a job, customers, funding, promotions or something else, we need to hoard as much as possible for ourselves because there is simply not enough to go around. This same theory also says that there are limited ways to achieve success, and that anyone who wishes to make it must follow the same path and prescription that others have done previously.

In contrast, this coach, through her word and deed, demonstrated to me a living illustration of what Covey labeled abundance theory, or AT. Abundance theory is a mindset that looks at each glass as half full (at least) and sees the world as offering endless opportunity. ... " 

Considering 'The Box'

In Innovation Excellence:    Working inside the box.  I like the idea.  When we build models there are always constraints that specify 'a box' we operate in.  You cannot ignore them.   Just need to consider when and where we apply them.