Professional Issues

Course Contacts

mcorey — Wed, 02 Aug 2023 11:54:43 +0000

Course Contacts mcorey Wed, 02/08/2023 - 12:54

Course queries should be directed to the lecturer, James Garforth (james.garforth@ed.ac.uk)

License

Course Information

mcorey — Thu, 27 Jul 2023 11:29:31 +0000

Course Information mcorey Thu, 27/07/2023 - 12:29

Outline of Topics

By week (estimated):

Introduction and Context
Responsibility
Power
Data Ownership
Bias and Fairness
The Human Element
Personal Attributes
Design Frameworks
Leadership
Teaching Others

Assessment

Marks for the course will be broken down as follows:

80% for two writing assignments (~1000 words) asking students to discuss topics covered by the course. One to be set mid semester, and one towards the end.
10% for contributing a peer review of another student's first writing assignment.
10% for tutorial group contributions (explained in more detail in the tutorial section).

Tutorials

Tutorial groups will meet with tutors every other week, starting in week 2. In the weeks in between, smaller subgroups of the tutorial will be tasked with preparing discussion points based on a reading or prompt provided to them, in preparation for the next full tutorial. Submission of these discussion points before the tutorial will be the criteria for the 10% discussion contribution mark.

Where to now?

This is a brief guide of where to look on the navigation panel to the right, depending on what you're after.

You want the timetable of events for the course, the lectures or readings? Go to Schedule.
You want to see what to prepare or submit for your next tutorial? Go to Tutorials.
You want to find the instructions for, or submit, your coursework? Go to Assessment.
You have questions about the course or want to discuss topics with your classmates? Go to Discussions (Piazza) via Learn.
You have a private question you need to ask the course organiser? Go to Course Contacts.

Course Outline

The course will be structured around professional and ethical behaviour, and the wider context in which technologies are developed and deployed. Beginning with the ACM Code of Ethics and Professional Conduct, the course will consider the wider context technologies are developed within and teach students to be considerate in their role as ICT professionals.

A standard breakdown of the course can be expected to be:
- Introduction (week 1 - gives an overview of the course)
- Responsibility (2 weeks, ACM principles 1.1, 1.2, 1.4): this will cover the responsibility of computing professionals. It will explore the notion of harms in the context of complex, multi-stakeholder situations, where benefit and harm are contested.
- Personal Attributes (2 weeks, ACM principles 1.3, 1.5, 1.6 and 1.7): this will cover personal attributes and why these are important by covering a range of situations that challenge professional integrity and work out how to respond to such challenges.
- Society (3 weeks, ACM principles 3.1, 3.2, 3.6 and 3.7) this will cover the obligations of computing professionals to recognise broader social requirements on their actions, particularly in areas where decisions involve the creation of new infrastructures that will underpin the delivery of public services or they are likely to be incorporated into widely used privately-owned platforms.
- Leadership (2 weeks, ACM principles 3.3, 3.4 and 3.5) will cover leadership amongst computing professionals, the obligations on leaders to ensure their leadership is fair and benefits those who are being led. This section will include a reflection on how these materials are taught.

The course will use articles and research from the social sciences, alongside short case studies drawn from contemporary situations that illustrate how knowledge of the decision-making context influences professional conduct and decision-making. Students will develop analytical skills to identify the critical influences on professionals in a range of real-world situations.

License

Tutorial 5 - Google AI Ethics

flittlet — Tue, 18 Jul 2023 14:57:37 +0000

Tutorial 5 - Google AI Ethics flittlet Tue, 18/07/2023 - 15:57

Summary

In December 2020, Timnit Gebru – computer scientist and co-lead of the Ethical Artificial Intelligence Team at Google – was fired after Google managers asked her to either retract her latest research paper or to remove her name from it, along with those of other members of the team who had contributed to the research. The paper in question, “On the Dangers of Stochastic Parrots: Can Language Models Be Too Big?”, was an overview of the environmental, financial, and social costs of large language models such as GPT-3. Google’s managers deemed the paper unpublishable but did not reveal how the internal review process they had come to this decision. Google claimed that Gebru had offered to resign and that her resignation had been accepted. In February 2021, Google fired the co-lead of the ethical AI team (and co-author of the paper), Margaret Mitchell, claiming that she had violated the company’s code of conduct. Google’s employees, who had previously protested the company’s collaboration with US Custom and Border Protection, the US Department of Defense, the Chinese state surveillance project Dragonfly, and had denounced Google’s handling of sexual harassment, demanded that Google provide clear guidelines on how research is reviewed and how research integrity is respected.

Core Readings

Hao, K. (2020, December 4). We read the paper that forced Timnit Gebru out of Google. Here’s what it says. The MIT Technology Review.

https://www.technologyreview.com/2020/12/04/1013294/google-ai-ethics-research-paper-forced-out-timnit-gebru/

Simonite, T. (2021, August 6). What really happened when Google ousted Timnit Gebru.

https://www.wired.com/story/google-timnit-gebru-ai-what-really-happened/

(2021, February 20). Google fires Margaret Mitchell, another top researcher on its AI ethics team.

https://www.theguardian.com/technology/2021/feb/19/google-fires-margaret-mitchell-ai-ethics-team

Prompts

Each subgroup should discuss one of these prompts (with the letter matching your subgroup letter) before the tutorial.

D) How would you define ‘ethics’ on an individual, organisational, and social level? What should be the relationship between these levels?

A) How does organisational culture affect the ways in which ethical research and development are conducted?
B) How can power structures within the organisation shape ethical research?

C) What do you think should be the relationship between an ethics team and the rest of the organisation?

License

Tutorial 4 - Robotic Care

flittlet — Tue, 18 Jul 2023 14:57:37 +0000

Tutorial 4 - Robotic Care flittlet Tue, 18/07/2023 - 15:57

Summary

Over the past decade, AI technologies and assistive robotics have become increasingly popular in nursing and healthcare settings. In particular, during the COVID-19 pandemic, socially assistive robots have been adopted to support elder care provision while promoting physical distance and reducing the spread of the virus. These robots are used to combat loneliness and inactivity through social or physical interaction in hospitals or elderly-care facilities. One such example is Paro, a therapeutic robot resembling a baby harp seal, designed and developed by Japanese manufacturer Intelligent Systems Co. to comfort elder patients, particularly those with dementia and other cognitive impairments. While studies have found that Paro encouraged engagement and social interactions, critics have raised concerns about the social and ethical implications of robotic care. In addition to concerns about the safety, privacy, and autonomy of those cared for, scholars have called attention to the impact of robotic care on care work and caregivers. Given the already unequal distribution of care work – often borne by low-paid or unpaid women, PoC, and migrants – robotic care risks further eroding the value of care labour. In contrast, scholars have argued that care is both a fundamental human capability and a critical public infrastructure.

Core Readings

Mateescu A., Eubanks V. (2021, June 3). ‘Care bots’ are on the rise and replacing human caregivers.

https://www.theguardian.com/us-news/2021/jun/03/care-bots-on-the-rise-elder-care

Vallor, S. (2011). Carebots and caregivers: Sustaining the ethical ideal of care in the twenty-first century. Philosophy & Technology, 24(3), 251-268.

https://discovered.ed.ac.uk/permalink/44UOE_INST/1viuo5v/cdi_gale_infotracacademiconefile_A373886519

Wang, X. (2021, August 26). A New AI Lexicon: Care. AI Now Institute.

https://medium.com/a-new-ai-lexicon/a-new-ai-lexicon-care-82c9031c98c4

Prompts

Each subgroup should discuss one of these prompts (with the letter matching your subgroup letter) before the tutorial.

C) How do you define care? What knowledge or experience do you consider to come up with this definition? What technical and non-technical solutions might you devise to improve care?
D) How would you define and promote trust in robotic care? And what level of trust would you expect from a system that cares for you?
A) What do you think is the impact of robotic care on the safety, autonomy, and privacy of those cared for?
B) How do you think robotic care changes the experience of caregivers and care labour? How might existing systems of care change over time with the introduction of robotic care?

License

Tutorial 3 - ChatGPT

flittlet — Tue, 18 Jul 2023 14:57:37 +0000

Tutorial 3 - ChatGPT flittlet Tue, 18/07/2023 - 15:57

Summary

Released in demo format on the 30th November 2022 and quickly picking up speed after that, ChatGPT is a large language model (LLM) produced by OpenAI, and represents a notable leap in both the interface and capabilities of conversational chatbots. While initially open for free access to anyone who created an account, later updates have been made accessible only to paid accounts. Other large tech companies have also started providing similar systems, such as Google's Bard, or customised interfaces, such as Microsoft's Bing Chat. This proliferation of powerful LLMs has had an impact across a a huge swathe of domains, and has raised a lot of both excitement and concern, over topics like misinformation or automated human experiences.

Core Readings

Lock, Samantha. (2022, December 5). What is AI chatbot phenomenon ChatGPT and could it replace humans? The Guardian.

https://www.theguardian.com/technology/2022/dec/05/what-is-ai-chatbot-phenomenon-chatgpt-and-could-it-replace-humans

Goetze, Trystan S. (2023, July 10). ChatGPT Reveals What We Value and What We Do Not. The Blog of the American Philosophical Association.

https://blog.apaonline.org/2023/07/10/chatgpt-reveals-what-we-value-and-what-we-do-not/

Alternative:

Heaven, Will D. (2023, March 3). The inside story of how ChatGPT was built from the people who made it. MIT Technology Review.

https://www.technologyreview.com/2023/03/03/1069311/inside-story-oral-history-how-chatgpt-built-openai/

Prompts

Each subgroup should discuss one of these prompts (with the letter matching your subgroup letter) before the tutorial.

B) ChatGPT has access to vast data from the internet in order to be able to process information about an enormous range of subjects, but it also required extensive data labelling work to try and identify harmful content. What problems, and for whom, can data usage on this scale cause?
C) Given that large language models have been accused of being well tuned for sounding convincing over being truthful, what consequences might the technology have for institutions built around reliability of information or other forms of trust?
D) Many sources have started recommending the use of ChatGPT or other “generative” models for automating tasks such as writing (CVs, essays, etc.), question answering, or art. What might be lost to the user (or indeed wider society over time) in using these tools in these ways?
A) The release of ChatGPT took a lot of people, including in some ways its developers, by surprise. In what ways do you think the manner of release of a technology like this influences its potential to cause harm?

License

Assessment

flittlet — Tue, 18 Jul 2023 14:57:37 +0000

Assessment flittlet Tue, 18/07/2023 - 15:57

License

Tutorial 2 - Cambridge Analytica

flittlet — Tue, 18 Jul 2023 14:57:36 +0000

Tutorial 2 - Cambridge Analytica flittlet Tue, 18/07/2023 - 15:57

Task Reminder

Ahead of the tutorial (week 4), you should:

Familiarise yourself with the case study through one or more of the readings.
Meet with your subgroup to discuss your tutorial prompt (approximately 50 minutes).
One member of your group should submit notes summarising this discussion on your tutorial group's discussion board (deadline is the next actual tutorial in week 4).

Don't feel like your discussion needs to stick tightly to answering the prompt, or indeed settle on an answer at all. Instead, use the prompt as a starting point and see what concepts or ideas it leads to. If further interesting questions come out of your discussion, note these down too. You don't need to mark it specifically on the submission, but you might want to make personal note of anything particularly interesting you'd like to discuss further with the whole tutorial.

Summary

In 2018, several media outlets reported that Cambridge Analytica, a British political consulting firm working on the Trump campaign, had accessed personal data of up to 87 million Facebook users for political advertising. The firm had harvested personal data through a Facebook app called "This Is Your Digital Life", consisting of a series of questions designed to create psychological profiles. In reality, the app also could access information on the user's friends network, and even direct messages, without the explicit permission of the user. The app developer, Aleksandr Kogan, had told Facebook that the data was collected exclusively for academic purposes. However, by sharing the data with Cambridge Analytica, he breached Facebook's terms of service. In 2018, the data breach was exposed by former Cambridge Analytica employee Christopher Wylie.

Core Readings

Cadwalladr C. & Graham-Harrison E. (2018a, March 17). Revealed: 50 million Facebook profiles harvested for Cambridge Analytica in major data breach. The Guardian.

https://www.theguardian.com/news/2018/mar/17/cambridge-analytica-facebook-influence-us-election

Cadwalladr C. & Graham-Harrison E. (2018b, March 17). How Cambridge Analytica turned Facebook ‘likes’ into a lucrative political tool. The Guardian.

https://www.theguardian.com/technology/2018/mar/17/facebook-cambridge-analytica-kogan-data-algorithm

Alternative:

Information Commissioner’s Office (2018). Investigation into the Use of Data Analytics in Political Campaigns.

https://ico.org.uk/media/action-weve-taken/2260271/investigation-into-the-use-of-data-analytics-in-political-campaigns-final-20181105.pdf?fbclid=IwAR3J6KMCI3eI74wawGnO-XOAvpKtwxRH_tFhPK1VlEh7oDJJ7ncR3oQb2pc

Prompts

Each subgroup should discuss one of these prompts (with the letter matching your subgroup letter) before the tutorial.

A) Traditionally, political campaigns have relied on survey data. How do you think Big Data has changed the way we conduct political campaigns, and what is its social impact?

B) What do you think are the social and ethical implications of ‘microtargeting’ (personalised targeted ads), and the assumptions behind its use? How could this impact existing power dynamics between governments, tech corporations, and citizens?

C) Whistleblowers can play a fundamental role in exposing organisational misconduct and unethical behaviour. What do you think is the role of organisations and tech professionals in promoting a culture of trust, transparency, and accountability?

D) What do you think is the role of other actors (governments, regulators, citizens, etc.) in promoting accountability in the tech sector?

License

Tutorial 1 - COMPAS

flittlet — Tue, 18 Jul 2023 14:57:36 +0000

Tutorial 1 - COMPAS flittlet Tue, 18/07/2023 - 15:57

Task

Ahead of the tutorial, familiarise yourself with the COMPAS case study through the readings below.

Summary

Correctional Offender Management Profiling for Alternative Sanctions (COMPAS) is a risk assessment algorithm developed by a company called Northpointe, Inc (now Equivant), designed to assess a defendant’s potential risk of recidivism – the likelihood that someone who has committed a crime will reoffend in the future. The algorithm, now used by several US jurisdictions, combines multiple sources of information, including a questionnaire filled out by defendants, to assign a recidivism risk score from 1 to 10 (however, the company has not disclosed how they modeled the algorithm, nor what kind of data they collected). The algorithm classifies offenders as low-risk (1-4), medium-risk (5-7), and high-risk (8-10). These results are then included in the defendant’s documentation and submitted to the sentencing judge. A study conducted in 2016 by ProPublica has shown that the algorithm is disproportionately biased against black defendants. In particular, the study found that, while COMPAS predicted recidivism in both groups with the same accuracy rate (60%), the algorithm treated black defendants unfairly with regards to incorrect scores. Indeed, black defendants were almost twice as likely to be labeled as high-risk, but did not actually re-offend, compared to their white counterparts (45% vs. 23%). Similarly, white defendants were almost twice as likely to be labeled as low risk when they did in fact re-offend, compared to their black counterparts (48% vs. 28%).

Core Readings

Angwin, J., Larson, J., Mattu, S., & Kirchner, L. (2016, May). Machine bias. ProPublica.

https://www.propublica.org/article/machine-bias-risk-assessments-in-criminal-sentencing

Corbett-Davies, S., Pierson, E., Feller, A., & Goel, S. (2016, October 17). A computer program used for bail and sentencing decisions was labeled biased against blacks. It’s actually not that clear. The Washington Post.

https://www.washingtonpost.com/news/monkey-cage/wp/2016/10/17/can-an-algorithm-be-racist-our-analysis-is-more-cautious-than-propublicas/

Submission

There is no submission required for this tutorial, but there will be for the subsequent ones.

Prompt

This tutorial will spend more time talking over the format and case study, but the following example prompt is provided as practice for how to think about prompts in later tutorials.

"How might this case study be an example of developers acting as though they have “absolute moral authority”, or an example of the limitations of computer science’s focus on quantitative methods?"

Advice:

If you don’t know how to approach this, try to think about the following:
- Do we need to define any terms?
- Are there other related questions that this brings up?
- Is there missing information we might need? (you are encouraged to search for more sources)
Don't feel like you should make a "new" point all the time. Respond to other people’s points by building on them or (respectfully) challenging them.
Remember you're not looking for a definitive answer here. The prompt is a starting point for discussion to branch off of (though try not to get off topic entirely).

License

Tutorial Format

flittlet — Tue, 18 Jul 2023 14:57:36 +0000

Tutorial Format flittlet Tue, 18/07/2023 - 15:57

Tutorial groups will meet with tutors every other week, starting in week 2. Tutorials are each based around a case study, chosen for its relevance to the material of the preceeding weeks, and some prompts intended to spark discussion.

In the weeks in between tutorials, smaller subgroups will be tasked with preparing discussion points based on one of the prompts (allocated to them). You should read about the case study, then meet with your subgroup (online or in person, your choice) to discuss, noting down the key points of that discussion.

Submission of discussion points before the tutorial will decide the 10% discussion contribution part of your mark for this course. These are marked at a very coarse level and it should be easy to get high marks as long as you engage with the task. A safe rule of thumb to aim for is at least a page of notes, and that they would be sufficient to spark conversations in the rest of your tutorial even if your subgroup were all sick.

To submit, go to the Discussion tab on the course Learn page, follow the link to Tutorial Discussions, and create a new response. To make things easier for your tutor, the response should start with a title that states which tutorial number and subgroup the submission is for, and you should "sign" it underneath with the names of the subgroup members who took part in the discussion.

Visit your group page on Learn to see the subgroup assignments (after week 2).

License

Week 7 - Personal Attributes

flittlet — Tue, 18 Jul 2023 14:57:35 +0000

Week 7 - Personal Attributes flittlet Tue, 18/07/2023 - 15:57

Video - Guest Lecture

The main material this week is a guest lecture from possibly the best person to talk about this topic, Professor Shannon Vallor, from 2021.

Bonus Lecture - Projects in Tech Ethics

The in-person lecture materials this week will be going over some examples of the kinds of projects you can work on as a computer scientist who is interested in ethics, responsibility or social good. This will also just generally talk about how the final year project process works, if you've not been told yet.

07_-_projects.pptx (1.81 MB)

Reading

Required - Virtues for Data Scientists

"Key Virtues for Data Science"

https://www.scu.edu/ethics/focus-areas/internet-ethics/resources/interviews-on-data-ethics/

A series of interviews with leading data ethics experts, concluding with a short piece pulling out some of the key virtues that data scientists (and other computer scientists) should endeavour to display. This conclusion piece is the core thing to read here. The interviews can be considered optional.

Optional - Refusal to Work

"When the implication is not to design (technology)"

https://dl.acm.org/doi/10.1145/1978942.1979275

Suggests that in some situations the right behaviour for someone designing something is to stop doing so. Presents a set of questions designers can ask themselves, to help make this decision. Links in to some readings earlier in the course that talk about the responsible behaviour sometimes being to disengage with making the system.

Optional - Codes of Practice

"ACM Code of Ethics"

https://www.acm.org/code-of-ethics

Hopefully most of you are familiar with this from SEPP, but the ACM Code of Ethics is a popular example of a code of practice intending to highlight the behaviour expected from an ethical computing professional. This course was built with it in mind.

Optional - Being Critical

"The Critical Engineering Manifesto"

https://criticalengineering.org/

Encourages a critical outlook from all kinds of engineers through 11 manifesto points illustrating the traits of the Critical Engineer.

License