Scavenger Hunt 7 Flashcards
What is AI?
A set of related technologies that seems to emulate human thinking and action
How do AI systems learn from prior and ongoing experience?
Provide it with a large learning database
Enormous amounts of inputs with correct answers
The machine then examines the questions and answers
Uses the data to create a model to receive future inputs and give answers
Arrive at its own conclusions through the use of the historical information
Why is AI becoming pervasive – what two key needs does it meet?
Usability
Access to vast amounts of data
Be able to explain one of the real-world applications for Artificial Intelligence presented in the lecture.
Processing Loan Applications:
- Your AI system will use big data of past loans to make decisions and will then create a model of this data
- The AI will self-adjust itself based on how correct it is or not.
What is a VPA, virtual personal assistant? VEA, virtual enterprise assistant? How are they different?
Virtual private assistant: owned and used by you, the employee, outside of the enterprises control and security, running on your hardware.
Virtual Enterprise Assistant: is used by the employee, but it is owned and controlled by the enterprise, they are able to integrate it tightly with enterprise systems, which your VPA cannot talk to.
What is meant by the term, “Explainable AI?” Why do we need AI to be explainable?
Need to understand how our systems work so the government will allow us to implement them.
Understand the model in terms of its, strengths, weaknesses, and how will it behave in the future
What is the supply situation for trained data scientists right now? How does AI Augmented Analytics help?
Trained data scientists are in critically short supply, we can’t get answers
Thanks to AI, we have broader access to examining big data and we do not need trained data scientists.
Humans only choose models that actually work, AI may not.
What is the goal of “continuous intelligence” systems?
Want systems that will sense a situation and respond to it real time
What is Generative AI? Where is it on the Hype Cycle?
AI which is able to generate information completely on its own; it is nearing the Peak of Inflated Expectations
With Large Language Model AI systems, how is supervised training different from unsupervised? Why is the choice of training dataset important?
Supervised
- Data is provided as a set of inputs + correct outputs
Unsupervised
- Data is unstructured
The training dataset is what allows the AI to generate its results
Be able to explain the following issues with Large Language Model Generative AI systems:
Hallucinations
- AI makes up information
The One Answer Problem
- The fact that in life, there is not one true answer sometimes
Jailbreaking
- Making the AI say things it isn’t supposed to say, such as how to make a bomb/weapons
What does the French phrase “La fin du monde” have to do with AI?
Refers to the “end of the world”
Refers to AI taking over the world and ultimately ending humans
What does Elon Musk think about AI?
He believes that “AI is the biggest risk that we face as a civilization.”
What impact does Erik Brynjolfsson think AI will have on society?
Sees it as the second machine age; we are running with the machines to combine our strengths with theirs to achieve more
“A vast and unprecedented boost to mental power should be a great boost to humanity, just as the earlier boost to physical power clearly was”
What is Gartner’s opinion of AI’s likely impact over the next 5 years?
Agree with Brynjolfsson
Cannot fully think like us, so they cannot fully replace us
“Through 2022, few jobs are fully replaceable, but most occupations will have at least some activities augmented by AI”
