Adaptability
Improvise and demonstrate flexibility in order to move forward (being able to deal with problems, pivot and improvise in order to move
... [Show More] forward)
What does adaptability look like?
being able to work with different kinds of people, making the most out of the new situation, Being able to pivot in the face of adversity, Being open-minded, Knowing there is not only one right way
Brainpower
Read More
0:00
/
0:00
Professional Attributes
Demonstrate responsible behaviors, time management, work ethic, positive attitude
Name 2 of the 10 executive functions
Planning
Organization
Task initiation
Time management
Working memory
Metacognition
Self-control
attention/focus
Flexibility
Perseverance
What are some professional attributes?
Responsible behaviors, Building a strong base by putting in the work even if its not being graded, Active engagement, preparation, and attendance
What are the 4 Gen AI models
Large language models
Art/image models
Video models
Audio models
What are the characteristics of these models?
Ubiquitous: it's everywhere in all areas of society
Undetectable: the use of gen AI is not reliably detectable
Transformative: gen AI will change how we live learn and teach
What is labelled data?
data that has a tag, helps you learn a pattern, you need a pattern so you can imitate it
How is labelled data used when training Gen AI models?
Used with supervised learning, you start with labelled data, then test it. Then unsupervised learnings gives unlabelled data which lets it find its own patters which can be harmful/wrong.
What are some concerns with labelled data?
The data is labelled by people, but concerns may arise about who labelled the data
How do Gen AI systems differ themselves from search engines?
search engines such as google are a collection of data that goes to work, and retrieves different webpages, organizes them based on relevance/popularity, while Gen AI is a collection of patterns that uses data to predict what should come next
Large language model
LLMs are similar to working with humans, because they are good at human tasks if given the right tools but can also be incorrect just like humans. They use deeper learning techniques and large data sets to summarize, generate and predict new content
Image Models
aim to create realistic and coherent images from scratch
Key features of LLM prompts
Role and goal → provides context from which to start → Example: "You are a friendly and professional Marketing Manager mentoring a university intern"
Give very clear instructions → Improves effectiveness/relevance of output → Example: "Create a digital marketing case study to improve the intern's ability to create appropriate marketing material for specific target markets"
Give examples and steps → provides better context (preferences, existing, knowledge/experience) and improves effectiveness of output → Example: "An example of effective marketing material for X target market is... Create a new scenario and ask the intern to describe the target market. Wait for an answer. Then ask the intern to propose the best medium. Wait for a reply... "
Caveat #1: There is no perfect prompt
Caveat #2: Prompting is just the start, how you interact with output is equally as important
methods of effectively interacting with, and evaluating large language model output
Be human and think critically
Push back on output
Challenge it, dont just accept it - evaluate it and be extremely active and engaged
Check the sources
Identify the parts you support and ask the AI to expand
Identify the parts you do not support and ask for editing
Reintegrate your own thinking
Recognize your LEADERSHIP role
Writing editor, selecting strengths and weaknesses, evaluating quality
Make multiple efforts, multiple attempts, multiple experiments (as you should do with all thinking/writing activities) with YOU giving feedback
Hallucinations
Inaccuracy in output
Opacity
- AI can perform well at some jobs but fails in many other circumstances
- The reason why it fails is unpredictable
- Not even developers know exactly what will work with the AI
- This is why we say AI has opacity and why the results that we receive aren't even clear to the users and companies that use AI
Alignment
"trying to make sure the behaviour os AI systems matches what we want and what we expect"
Top-Down alignment
designers explicitly specify the values and ethical principles for AI to follows
Bottom-Up Alignment
reverse-engineer human values from data and build AI systems aligned with those values
Values and limitations of ethical uses of Gen AI on employment
Positive impact on employment
Add jobs
Increase incomes
Increase productivity and growth
May add trillions of dollars in value to global economy
May enable labour productivity growth [Show Less]