Tuesday, November 26, 2019

Police the Machines

Machine harm, like any other vice, requires no motive outside itself. It only requires opportunity ............... Emmanuel.Sodipo



Technology is now deeply intertwined with our socio-technical systems at all levels. Software now constrains behavior with an efficiency that no law can match. Technology is literally creating a new world and policies can’t keep up.

Getting it wrong has become increasingly catastrophic and a growing constituency is starting to hold reckless tech businesses accountable.

Consider artificial intelligence (AI), the technology that has the potential to augment human decision making with something more scalable and faster. But also has the potential to entrench bias and codify undesirable acts. It can be hacked in new ways with new capabilities to disrupt and harm.

Cognitive Architectures:
These models simulate human cognition, including perception, memory, reasoning, and problem-solving. Examples include ACT-R (Adaptive Control of Thought-Rational) and Soar.

Evolutionary Algorithms:
Inspired by natural evolution, these algorithms optimize solutions by evolving a population of potential solutions over generations. Genetic algorithms and genetic programming fall into this category.

 Other technologies we should police are:
  • Critical Cybersecurity vulnerabilities 
  • Role of social media platforms in propaganda 
  • Bioengineering such as genome editing and synthetic biology.
  • Robotics as a common consumer technology

Ethics owners and policy makers must navigate between avoiding measurable downside risk and promoting the upside benefits. Arguing against releasing a product before it undergoes additional testing for racial or gender bias is one thing. Arguing that a more extensive test will lead to greater sales is another thing.

Both are important but one sits with compliance the other with marketing. 

We should aim to achieve a robust process rather than a substantive outcome.

See You at the Top


No comments: