BUS 121W - Week 10

Class - Mar. 17, 2026 & March 19, 2026

Amplify income and wealth inequality at individual and macro level
- Will development accentuate power of mammoth tech corporations
Reduce productivity gains
- AI-generated 'workslop' puts burden on receiver of output to revise and improve
  - Wastes time and images of how colleagues see each other
  - Respect for work culture lost

Increase productivity in the coming years and decades
- Very similar adoption pattern as electricity
- Co-invention will drive real economic transformation
  - And hopefullly result in innovative breakthroughs
- Experimentation needs to start now because it learns from use
  - Companies want AI to learn about their data, systems, workflows
- Adoption path may not be linear but long term, every sector will be impacted
Shifts where new graduates work
- Early data show declines in entry level, AI-exposed fields but OVERALL employment rose and wages did not fall
- Young workers may be shifting to startups
Adjust skills that drive success in knowledge economy
- AI is improving rapidly at prediction but cannot perform judgement (this is actually getting better)
- Competition moves 'upstream'
- Judgement developed through experience making trade-offs = superpower
Expand demand for services
- Who works in organizations may change but if cost of cognitive tasks falls, demand may increase

Increases educational divide
- Billions of people are without access to information and communication technology
- Even with technology access, digital literacy is not evenly shared (quality of prompt makes difference to how well a tool can help someone)
- Some versions of Gen AI require payment, creating financial barrier
Increase in academic misconduct (LOL)
Decrease in deep learning and competency development
Decrease in authentic research and authorship
Increases access to quality education and support democratization of knowledge
- May offer scalable, personalized higher quality tutoring
  - Can capably respond to prompts across a myriad of areas with 3Cs
- may offer teaching assistant and teaching resources for every teacher
  - Can diagnose gaps in student learning
  - Can provide timely feedback to students
- Support writing tasks
Elevate critical analysis and rigor in assessment of knowledge
Enhance human intelligence, human potential, and human purpose

Accelerate environmental problems
- Significant energy required to train Gen AI models and deploy them in real-world applications
  - Pace of growth in data centers requires most of the electricity to come from fossil fuel-based power plants
  - Ubiquity of use and increasing complexity of future models may mean interference uses more energy than training (HOWEVER: individual consumption is relative to use of other electronics)

NOTE: must consider both operational carbon emissions and embodied carbon emissions
- Embodied carbon emissions are all about how data centers are built in the first place
Reduce operational carbon emissions
- 'Turning down' GPUs can reduce consumption to 30% without significantly impacting performance
- Reduce precision of computing hardware
- Stop training process slightly earlier (half the electricity used for training is spent to get last 2-3% points in accuracy)
Leverage efficiency improvements
- Computations per chip continues to improve
Maximize energy savings
- Split/delay operations to different times and sources
Use AI to develop solutions
- Optimize prediction of solar and wind energy generation
- Perform predictive maintenance and fault detection for green energy infrastructure

Increase potential for proliferation of misinformation and disinformation
- We believe what is false
- We don't believe what is true

Perpetuate outdated gender norms and stereotypes
Reinforce and perpetuate social bias and inequities
- Training data
  - Scrapes the internet and gets data like from Reddit
- Missing data
- lack of diverse AI developers
Create potential for 'diversity-washing'
Make data driven decisions that reduce human bias

Increases ethical issues regarding consent, copyright and intellectual property
- No AI company can explicitly state exactly where they get all their information?
- Creatives may lose control over rights and compensation
Increase concerns regarding control over personal data and privacy

Conducted by Future of Life institute, a non-profit dedicated to helping humanity educe negative outcomes from powerful technologies
None of the current activity provides a clear, quanititative guarantee of safety