Get Started. It's Free
or sign up with your email address
Rocket clouds
Ethics in AI by Mind Map: Ethics in AI

1. Accuracy

1.1. Ethical ramifications

1.1.1. Precision versus recall trade-off - false positives and false negatives

2. What are others thinking/doing?

2.1. Google's AI principles

2.2. Vodafone

2.3. Facebook

2.4. MIT

2.5. DeepMind's ethical principles

3. Societal impacts

3.1. Labour market

3.1.1. Working conditions

3.2. Income inequality

3.3. Reliance on technology

3.3.1. Inability to disconnect

3.3.2. Sacrificing privacy for connection Inability to understand how own personal data is used

3.3.3. People's work and lives are retrofitted to serve machines. Ref: Prof Stephanie Dick

4. Other terms

4.1. Ethically aware algorithims

4.2. Human-centred machine learning

4.3. Inclusive machine learning and AI

4.4. Mindful AI

5. Human judgement in the data science progress

5.1. People choose where the data comes from, and why they think the selected examples are representative.

5.2. People define what determines success, and further, what evaluation metrics to use in measuring whether or not the model is working as intended.

5.3. People are affected by the results.

6. Types of licences

6.1. Unrestrictive

6.2. Copyleft

7. Biases

7.1. Cognitive

7.1.1. Social

7.2. Statistical

7.3. Algorithmic

8. Approaches to principles

8.1. Applying systems thinking

8.2. Do no harm or only do good?

8.3. Belmont human subject research principles

8.4. Thinking about regulation of algorithms

9. When decisions are made by machines

9.1. Designing for transparency

9.2. Designing for a system of augmentation rather than replacement of decisions

9.3. Explainability of decisions is an important principles here

9.3.1. What kind of explanations are important? Counterfactuals as explanations. Ref: Alan Turing Institute Brent Mittelstadt and Prof Sandra Watcher

10. How machines and human interact

10.1. AI machines learn from humans in the case of reinforcement learning and deep learning - imitation learning

10.1.1. Deep learning is essentially an autonomous, self-teaching system in which you use existing data to train algorithms to find patterns and then use that to make predictions about new data.

10.1.2. Reinforcement learning is an autonomous, self-teaching system that essentially learns by trial and error. It performs actions with the aim of maximizing rewards, or in other words, it is learning by doing in order to achieve the best outcomes.

10.1.3. This is where we should focus our attention towards the dangers of society biases creeping into the algorithms

10.2. AI machines can also learn efficiently without learning from the behaviour of humans, but in the case of reinforcement learning they learn based on trial and error within a system of rules. Ref: Thore Graepal (DeepMind)

10.2.1. This seems to be an area of less ethical concern in terms of unintended consequences

10.3. Humans can learn from machines. You can use AI Models to examine complex human systems, e.g. the property market - and then can more easily identify points of sensitive intervention. Ref: Eric Beinhocker

10.4. Anthropormorphisms and Robomorphism

10.4.1. Robomorphism: We adapting ourselves to machines. We have to change our behaviour to fit with the technology. Retrofitting humans to technology, e.g. CVs, passport machines. Ref: Beth Singler.

10.5. Affective computing:

10.5.1. Affective robots and machines can teach people with autism how to read facial emotions. Ref: Maja Pantic, Imperial College London.

10.5.2. Affective machines can can read facial emotions can detect depression. Ref: Maja Pantic, Imperial College London

10.6. 'We need to nurture algorithms in the small way we nurture children'

10.7. Automation: What human abilities should be automated and which should not?

10.7.1. The way we automate and partition cognitive tasks is part of a long history of valuing and de-valuing forms of labour. What groups are suffering as a result of the re-distribution of labour? What parts of labour being disciplined to become closer to machines? Ref: Stephanie Dick Penn

10.8. It's a symbiotic relationship

11. Safety of AI algorithms

11.1. DeepFakes and disinformation

12. Types of AI

12.1. Machine learning

12.1.1. Supervised learning

12.1.2. Unsupervised learning Deep learning neural network

12.1.3. Reinforcement learning Multi-Agent learning

12.2. Natural language processing

12.2.1. Word embeddings Ethical issue: Man is to Computer Programmer as Woman is to Homemaker? Debiasing Word Embeddings

13. Explainability

13.1. Black box explanations

13.1.1. Explanations based on counterfactuals: Watcher et all 2019 Advantages: Accessible to the lay person and they can challenge the decision based on explanation Limitation: Lacks interpretability - not enough for full transparency