AI for Everyone Quizzes
ML programs can help: ・Automate resume screening ・Automate lead sorting in sales ・Customize product recommendations ・Automate visual inspection in a manufacturing line
All of the above
What are the key steps to a smart speaker function?
Trigger word detection -> speech recognition -> intent recognition -> command execution
According to the AI Transformation Playbook, broad AI training needs to be provided not only to engineers but also to executives/senior business leaders and to leaders of divisions working on AI projects. ・True ・False
True
Say you want to use ML to help your sales team with automatic lead sorting. Input A (a sales prospect) and output B (whether your sales team should prioritize them). The 3 steps of the workflow are:
1. Collect data with both A and B 2. Train an ML system to input A and output B 3. Deploy a trained model and get data back from users
Which of these terms best describes the type of AI used in today's email spam filters, speech recognition, and other specific applications? ・Artificial General Intelligence (AGI) ・Artificial Narrow Intelligence (ANI)
Artificial Narrow Intelligence (ANI)
What is the first step in the AI Transformation Playbook for helping your company become good at AI? ・Provide broad AI training ・Develop an AI strategy ・Execute pilot projects to gain momentum ・Build an in-house AI team
Execute pilot projects to gain momentum
Because smart speakers can carry out multiple functions (such as tell a joke, play music) it is an example of Artificial General Intelligence (AGI). ・True ・False
False
The only way to acquire data for a supervised learning algorithm is to manually label it. (given the input A, ask a human to provide B) ・True ・False
False
Unless you have a huge dataset ("Big Data"), it is generally not worth attempting ML or DS projects on your problem. ・True ・False
False
What are the jobs that AI is most likely to displace over the next several years? ・Jobs that comprise primarily of non-routine, non-repetitive work ・Most jobs involving office work (white collar jobs) ・All jobs will be displaced ・Jobs that comprise primarily routine, repetitive work
Jobs that comprise primarily routine, repetitive work
Say you are building an AI system to help make diagnoses from X-ray scans. Which of the following statements about explainability of AI do you agree with? ・Lack of explainability can hamper users' willingness to trust and adopt an AI system. ・Most AI systems are highly explainable, meaning that it's easy for a doctor to figure out why an AI system gave a particular diagnosis. ・AI systems are intrinsically "black box" and cannot give any explanation for their outputs. ・Explainability is usually achieved through building a chatbot to talk to the user to explain its outputs.
Lack of explainability can hamper users' willingness to trust and adopt an AI system.
Suppose you are building a trigger word detection system, and want to hire someone to build a system to map from Inputs A (audio clip) to Outputs B (whether the trigger word was said), using existing AI technology. Out of the list below, which of the following hires would be most suitable for writing this software? ・AI Product Manager ・Data Engineer ・ML Engineer ・ML researcher
ML Engineer
What is the Goldilocks Rule of AI? ・One should allocate many resources to defend the world from giant killer robots ・One shouldn't be too optimistic or too pessimistic about AI technology ・AI's technology will continue to grow and can only benefit society ・An AI winter is coming
One shouldn't be too optimistic or too pessimistic about AI technology
Which of these statements regarding data acquisition do you agree with? ・Some types of data are more valuable than others; working with an AI team can help you figure out what data to acquire. ・It doesn't help to give data to an AI team, because they can always produce whatever they need by themselves. ・It doesn't matter how data is acquired. The more data, the better. ・Only structured data is valuable; AI cannot process unstructured data.
Some types of data are more valuable than others; working with an AI team can help you figure out what data to acquire.
Of the following options, which is the most important trait of your first pilot project? ・Succeed and show traction within 6-10 months ・Drive extremely high value for the business ・Be executed by an in-house team ・None of the above
Succeed and show traction within 6-10 months
What do you call the commonly used AI technology for learning input (A) to output (B) mappings? ・Supervised learning ・Unsupervised learning ・Reinforcement learning ・Artificial General Intelligence
Supervised learning
If a developing economy has a strong and thriving coffee bean manufacturing industry (or some other vertical industry), then it has an advantage in applying AI to coffee bean manufacturing (or other vertical industry). ・True ・False
True
ML is an "iterative" process, meaning that an AI team often has to try many ideas before coming up with something that's good enough, rather than have the first thing they try work. ・True ・False
True
Say you want to input a picture of a person's face (A), and output whether or not they are smiling (B). Because this is a task that most humans can do in less than 1 second, supervised learning can probably learn this A-to-B mapping. ・True ・False
True
Using current AI technology, if an ML system learns only from text that is completely neutral and does not reflect any gender biases, then we would expect it to exhibit no, or at most minimal, gender bias. ・True ・False
True
Using current AI technology, if an ML system learns from text that reflects unhealthy biases/stereotypes, then the resulting AI software may also exhibit similarly unhealthy biases/stereotypes. ・True ・False
Ture
Why is developing an AI strategy NOT the first step in the AI Transformation Playbook? ・When transforming a company into an AI company, one does not need a strategy, therefore it can't be the first step. ・The strategy should be to use the Virtuous Circle of AI, which comes after building a product. ・Without having some practical AI experience and knowing what it feels like to build an AI project, a company usually does not know enough to formulate a sound strategy. ・There is no reason. Developing an AI strategy IS the first step in the AI Transformation Playbook.
Without having some practical AI experience and knowing what it feels like to build an AI project, a company usually does not know enough to formulate a sound strategy.
Suppose you run a website that sells cat food. Which of these might be a good result from a DS project? ・A slide deck presenting a plan on how to modify pricing in order to improve sales. ・A neural network that closely mimics how cats' brains work. ・Insights into how to market cat food more effectively, depending on the breed of cat. ・A large dataset of images labeled as "Cat" and "Not Cat"
・A slide deck presenting a plan on how to modify pricing in order to improve sales. ・Insights into how to market cat food more effectively, depending on the breed of cat.
What are the current limitations of AI technology? ・AI technology can be biased ・There are no limitations to AI technology ・Explainability is hard ・AI technology is susceptible to adversarial attacks ・AI technology can discriminate
・AI technology can be biased ・Explainability is hard ・AI technology is susceptible to adversarial attacks ・AI technology can discriminate
Say you are building a smart speaker, and want to accumulate data for your product having many users. Which of these represents the "Virtuous circle of AI" for this product? ・(more users of speaker -> more user data -> better smart speaker) ・(more user data -> more users of speaker -> better smart speaker) ・(more commands supported -> more users of speaker -> better smart speaker) ・(more users of speaker -> more commands supported -> better smart speaker)
・(more users of speaker -> more user data -> better smart speaker)
You want to use supervised learning to build a speech recognition system. The figure above suggests that in order for a neural network (deep learning) to achieve the best performance, you would ideally use: ・A large dataset (of audio files corresponding text transcript) ・A small dataset (of audio files corresponding text transcript) ・A large neural network ・A small neural network
・A large dataset (of audio files corresponding text transcript) ・A large neural network
You run a company that manufactures scooters. Which of the following examples of unstructured data? ・The number of scooters sold per week over the past year. ・The maximum speed of each of your scooters. ・Audio files of the engine sound of your scooters. ・Pictures of your scooters.
・Audio files of the engine sound of your scooters. ・Pictures of your scooters.
Which of these statements about "business diligence" do you agree with? ・Business diligence applies only if you are launching new product lines or businesses. ・Business diligence is the process of ensuring that the AI technology, if it is built, is valuable for your business. ・Business diligence can typically be completed in less than a day. ・Business diligence is the process of ensuring that the envisioned AI technology is feasible.
・Business diligence is the process of ensuring that the AI technology, if it is built, is valuable for your business.
Key steps of a DS project?
・Collect data ・Analyze data ・Suggest hypothesis or actions
Which of the following statements do you agree with? ・DL is a type of ML (e.g. all DL algorithms are ML algorithms) ・AI is a type of DL (e.g. all AI algorithms are DL algorithms) ・The terms "ML" and "DS" are used almost interchangeably. ・The terms "DL" and "neural network" are used almost interchangeably.
・DL is a type of ML (e.g. all DL algorithms are ML algorithms) ・The terms "DL" and "neural network" are used almost interchangeably.
Which of the following are AI pitfalls to avoid? ・Expecting an AI based projects to work the first time ・Expecting AI to solve everything ・Expecting traditional planning processes to apply without changes ・Pairing engineering talent with business talents to identify feasible and valuable projects
・Expecting an AI based projects to work the first time ・Expecting AI to solve everything ・Expecting traditional planning processes to apply without changes
For your automated resume screening application, you are now providing a Test Set to the AI team. Which of the following statements about the Test Set are true? ・It will be used by the AI team to evaluate the performance of the algorithm. ・It should give examples of the input A (resume) but not necessarily the desired output B ( whether to move forward with a candidate). ・It should give examples of both the input A (resume) and the desired output B (whether to move forward with a candidate). ・The Test Set should ideally be identical to the Training Set
・It will be used by the AI team to evaluate the performance of the algorithm. ・It should give examples of both the input A (resume) and the desired output B (whether to move forward with a candidate).
You want to use supervised learning for automated resume screening, as in the example above. Which of the following statements about the Training Set are true? ・It will be used by the AI team to train the supervised learning algorithm. ・It should give examples of the input A (resume) but not necessarily the desired output B ( whether to move forward with a candidate). ・The Training Set and the Test Set can be the same dataset. ・It should give examples of both the input A (resume) and the desired output B (whether to move forward with a candidate).
・It will be used by the AI team to train the supervised learning algorithm. ・It should give examples of both the input A (resume) and the desired output B (whether to move forward with a candidate).
Say you want to build an AI system to help recruiters with automated resume screening. Which of these steps might be involved in "technical diligence" process? ・Making sure that an AI system can meet the desired performance ・Ensuring that this is valuable for your business (e.g. estimating the project ROI) ・Making sure you can get enough data for this project ・Defining an engineering timeline
・Making sure that an AI system can meet the desired performance ・Making sure you can get enough data for this project ・Defining an engineering timeline
Applications of Supervised Learning
・Spam filtering ・Speech recognition ・Machine Translation ・Online advertising ・Self-driving car ・Visual Inspection
What do AI companies do well?
・Strategic data acquisition ・Invest in unified data warehouses ・Spot automation opportunities
Examples of adversarial attacks on an AI system?
・Subtly modifying an audio clip to make a speech recognition system think someone said "Yes, authorized" when they actually said "No, reject." ・Adding a sticker to a stop sign to make an AI system fail to detect it. ・Subtly changing an image to make an AI system mistakenly recognize a dog as a cat.
What are good practices for addressing bias in AI?
・Using more inclusive/less biased data ・Technical solution such as "zeroing out" bias ・Systematic auditing processes to check for bias
What are the reasons that it's often unrealistic to expect an ML system to be 100% accurate?
・You might not have enough data ・Data can be mislabeled ・Data can be ambiguous