Lecture 5 Flashcards
what does abstraction error mean?
treating fairness and justice as concepts separable from their inherent connection to the social context.
What is the portability trap and what does it refer to?
The portability trap is the failure to understand how repurposing algorithmic solutions designed for one social context may be misleading, inaccurate, or otherwise do harm when applied to a different context.
What are two problems that can arise due to the formalism trap?
1) There is no single definition of concepts like fairness that will satisfy all stakeholders.
2) Formalizations are simplifications that may not apply across different social contexts.
What is the ripple effect trap and how can it destabilize existing systems?
The ripple effect trap is the failure to understand how the insertion of technology into an existing social system changes the behaviours and embedded values of the pre-existing system.
It refers to technologies inspiring reactivity that may destabilize values, incentives and structures such that the tool no longer solves the intended problems.
What is the solutionism trap and what does it caution?
The solutionism trap is the failure to recognize that the best solution to a problem may not involve technology.
It cautions against implementing new technology without first considering if there are non-technical solutions or factors involved.
What is the framing trap and how can it be avoided?
The framing trap is the failure to model the entire system a social criterion will be enforced over.
It can be avoided by adopting a “heterogeneous engineering” approach that draws analytical boundaries around both technical and human components of the sociotechnical system.
what are the three themes of possible consequences when AI systems reach work and organisation?
- changes to knowledge work and expertise.
- new forms of control.
- changes to knowledge gathering and learning.
what are the changes in knowledge work and expertise with the introduction of AI?
- Tasks that involve expertise or tacit knowledge, where part of the knowledge cannot be directly explained, are increasingly being performed by AI systems.
- For professions where extensive training and practice are required to develop expertise, it can be difficult to maintain that level of expertise if AI systems start performing part of the work. This is because professionals would no longer be getting as many hours of hands-on experience
- While AI can take over more routine tasks, freeing up experts to focus on more complex work requiring their expertise, there is a concern that over-reliance on AI could reduce opportunities for professionals to develop and maintain their skills
- It is unclear who would be responsible for the consequences if AI systems are used to perform parts of expertise-based work traditionally done by humans, like medical diagnoses
what is meant by new forms of control?
new forms of control refer to how AI-enabled data collection and analysis can shift the balance of power between organizational oversight and employee autonomy and freedom in their work.
- AI systems can provide concrete predictions and expectations about employee output based on complex statistical calculations of employee data, which limit employee freedom and autonomy in how work is done.
- Organizations have more data available about internal work processes and employee activities due to use of AI. This gives them more visibility into who and what provides value. However, it can also feel more restrictive and limit employee freedom if every action is recorded and reviewed.
- Rigid systems for employees to report on their work using data can disrupt the balance between professional freedom and organizational control. It may even negatively impact creativity and flexibility if employees feel they need to structure their work according to what is recorded in data systems.
- There is a challenge for organizations in determining when data collection and use enables work versus becoming a form of “false control” that hinders work processes
what is meant by changes to knowledge gathering and learning?
while AI aims to enhance knowledge gathering and learning, its black box nature and narrow capabilities also introduce challenges and uncertainties around organizational understanding and insight.
- AI systems make it easier to visualize work patterns and choices that impact organizational performance. This can benefit making decisions in a more objective, data-driven manner.
- When used to find and adjust work patterns, AI is said to improve an organization’s ability to gather knowledge and learn how to improve. However, the “black box” nature of many AI systems means people do not understand their exact functioning. This limits organizational learning, as the reasons for outcomes may not be clear.
- Narrow AI systems today can only perform predefined tasks and cannot provide unexpected insights. This may constrain knowledge gathering compared to human experts.
- There is a question around what type and depth of knowledge AI systems can truly offer organizations, given their current limitations
what are some challenges associated with anticipating indirect changes to work from introducing AI systems?
It is difficult because changes to one aspect of work can influence other aspects in unexpected ways due to interconnected nature of knowledge work.
What perspective is needed to better anticipate indirect changes to work?
A sociotechnical perspective that considers the entire work process and how tasks are embedded within it, not just specific tasks.
How does the text recommend organizations approach changes in work from AI?
Continuously monitor work processes over time to determine how work is changing and if changes are as intended/meaningful.
how can AI systems be used to augment work?
By automating routine processes to free up time for more meaningful work like personal contact.
What type of forms between humans and AI systems are being developed?
- Collaboration between users and AI in hybrid development leads to AI being supportive, serving as assistants, coaches, and colleagues in various professions.
