Holistic AI Ethics by Ben Byford

We seek to promote thoughtful collaboration and open dialogue about the potential subtle and salient influences of AI on people and society.

Good AI offers great potential for promoting the public good for example in the realms of education, housing, public health and sustainability

There are concerns about algorithms’ opacity the code of algorithms may be unviewable in systems that are proprietary or outsourced, even if viewable the code may be essentially uncheckable, if it’s highly complex where the code continuously changes based on live data or where the use of neural networks means that there is no single ‘point of decision making’ to view before data may go through...

Does your company communicate clearly, honestly and directly about any potential risks of the product or service you are providing what does it communicate?

Holistic AI Ethics by Ben Byford

We seek to promote thoughtful collaboration and open dialogue about the potential subtle and salient influences of AI on people and society.

Good AI offers great potential for promoting the public good for example in the realms of education, housing, public health and sustainability

There are concerns about algorithms’ opacity the code of algorithms may be unviewable in systems that are proprietary or outsourced, even if viewable the code may be essentially uncheckable, if it’s highly complex where the code continuously changes based on live data or where the use of neural networks means that there is no single ‘point of decision making’ to view before data may go through...

Does your company communicate clearly, honestly and directly about any potential risks of the product or service you are providing what does it communicate?

Who am I?

What is Ethics?

Holistic AI ethics

AI Ethical guidelines

What we need to do now

Questions

What is Ethics?

wikipedia

 

Ethics or moral philosophy is a branch of philosophy that involves systematizing, defending, and recommending concepts of right and wrong conduct.

Ethical = Good

Ethical ≠ Good

...systematizing conduct.

 

Ethics is an agreed
code of conduct for societies.
The framework

 

Morals are the framework's parts,
it's good and bad decisions

(This is a working definition for this talk)

In what way do machines participate in ethical questions?

Why AI Ethics?

“It is impossible to work in information technology without also engaging in social engineering.”

 

Jaron Lanier, You Are Not a Gadget

"…ethics must become a central piece for any individual or organisations developing AI."

 

DigitalCatapult

Machine Intelligence Garage Ethics Committee
chaired by Luciano Floridi

 

Why Holistic AI ethics?

Tech Ethics

Data Ethics

AI ethics

RoboEthics

Machine Ethics

AGI / theory of mind

Where do these decisions get made?

 

Govenments

Business

Design

Researchers

Technology

Data

AI in Warfare, regulation of malpractise, best practise / investing in research, bad actors and security, democratisation of tools, education or get left behind, using other companies algorithms are they ok, company acquisition, team diversity, what does success look like, impact on currently held freedoms e.g free speech, democracy, monopolies of algorithmic power, job displacement,
who is accountable, what is the visibility of that account, secrecy, how to implement and where, unintended consequences, can the users interpret the system, is it exploitative, false positives and true negatives, is the system designed dogmatically, what behaviours is the system promoting, discriminatory, how can the algorithm fail, does the system change over time and does it need to be repeatedly tested for changes, Is the output explainable - does it need to be, what bias do you have in the data,
do you have enough data to make descisions, Is the data reliable, is it appropriate, proper consents, Inference of anonymised personal data, what score / threshold is good for your domain?

“Perhaps interacting with an ethical robot might someday even inspire us to behave more ethically ourselves.”

 

Anderson, Anderson 2012

Ethical
frameworks, guidelines, manifestos, declarations, principles practises, questions, recommendations, toolkits...

Opinions, doctrine, tenets, obligations, laws

Good, meh, common, miss ng

The good

 

  • Avoid creating or reinforcing unfair bias
  • Only collect what is needed
  • Monitor after release of systems its effectiveness but also its impact on societies
  • Do some of the benefits of your solution include democratising AI knowledge?
  • Can you give others access to your data and models to help charities and NGOs?
  • Build ethical thinking into organisation culture

The good

 

IBM - everyday ethics

https://www.ibm.com/watson/assets/duo/pdf/everydayethics.pdf

 

Digital catapult - machine intelligence garage, Luciano Floridi

https://www.migarage.ai/ethics-framework/

The common

 

  • Transparency of where an algorithm is being deployed
  • Diverse teams
  • Data consent

The meh

 

  • User’s should be in control of their data - GDPR
  • Respect human rights
  • Dedicated insurances

 

The bizarre

  • Take action licenses

The miss ng

 

  • Acceptance and handling of unintended Consequences
  • Specifics of the technology
  • Specifics of a generalised ethically framework based on logical reasoning in relation to the technology
  • No recommendations for legislation
    (YET but its coming)

“...its just not good enough to publish your ethics principles”

 

Projects by IF

OECD AI Principles

 

www.oecd.org/going-digital/ai/principles/

  • AI should benefit people and the planet by driving inclusive growth, sustainable development and well-being.

  • AI systems should be designed in a way that respects the rule of law, human rights, democratic values and diversity, AND they should include appropriate safeguards – for example, enabling human intervention where necessary – to ensure a fair and just society.

  • There should be transparency and responsible disclosure around AI systems to ensure that people understand AI-based outcomes and can challenge them.

  • AI systems must function in a robust, secure and safe way throughout their life cycles and potential risks should be continually assessed and managed.

  • Organisations and individuals developing, deploying or operating AI systems should be held accountable for their proper functioning in line with the above principles.

  • respect humans

  • respect the planet

  • safeguards

  • transparent

  • robust

  • accountable

What we need to do now?

ALOT

 

Govenments -> Regulation(?), Signpost good practise, help us decide on basic principles for tech ethics and more.

 

Businesses -> hold themselves accountable,
create services in societies interest

 

Designers / developers / data scientists -> Collaborate on ethical issues, whistleblow where human right obligations are not being met

 

Research -> continued research on Programmed ethical agents (machine ethics) which needs to be more transparent and less click baity (terminator / your car is going to kill you)

 

Much more...

Extra credits:

Recommendations
for your projects

1. Hire me: www.ethicalby.design

 

 

1. Hire me: www.ethicalby.design

 

 

2. Create an AI governance group:

designers, users, directors, data scientists, allow these people to "own" the process of scrutony for your projects... They're like AI fire wardens.

 

3. Risk assess your service / product against ethical issues, use this process to put processes and changes into place.

 

4. Revisit your AI processes, reevalute them and keep up to date with new AI issues.

 

5. Publicily declare your intent, processes and updates... you don't have to give up your secret source just give us as much as possible to be able to see your doing well.

"I'm just a
a developer,
a researcher,

a business"

“Integrity without knowledge is weak and useless, and knowledge without integrity is dangerous and dreadful.”

 

Samuel Johnson

Thank you

Learn more here: www.machine-ethics.net

Hire me here: www.ethicalby.design

@BenByford

Yes there's more

NOTES

 

  • Finding bias isn't the same as acting on it
  • Users - humanise your users by calling them people
  • “Be worthy of trust” - sounds virtuous, but how?
  • Badges for ethical & technical understandings
  • Fight against abstraction
  • ...TBA
Made with Slides.com