AI 2.2

Why is LinkedIn a useful resource for data professionals? Select all that apply.

• It’s a way to follow industry trends.

• It’s a way to learn from thought-leaders.

• It can help you to stay engaged with the global data analytics community.

• You can give yourself accomplishments regardless of whether or not they are true, because nobody verifies what you say.

For which of the following use cases is it appropriate for a data professional to use a large-language model (LLM) like Gemini or ChatGPT? Select all that apply.

• To help bypass security systems that slow down your retrieval of data

• To provide an outline to use as a template for a report

• To help generate code to visualize data

• To brainstorm ideas for how to approach a problem

What is an important measure for organizations to counter potential human bias that can be imprinted within AI-driven work?

• Focusing AI development solely on tasks that do not involve human-related data to avoid bias

• Ensuring that all AI models are developed and maintained by a small, highly specialized group of AI experts

• Exclusively using AI systems that have been trained on globally standardized datasets

• Building diverse teams of professionals from different backgrounds and life experiences

Which of the following would be an appropriate way for a data professional to use AI? Select all that apply.

• A statistician uses an LLM to help break down a difficult statistical concept into simple steps and analogies so he can effectively communicate the concept to an executive without expertise in this area.

• A data analyst needs to write a complex regular expression that will identify and extract certain segments of text from a list of user-submitted product reviews, so they ask an LLM for help writing the code and verifying that it works as intended.

• A data scientist for an insurance company uses an LLM to create a machine learning model that automatically sets each customer’s insurance rates without the need for human oversight.

• A data analyst wants to add new information to an existing spreadsheet, so they add the column name and ask an LLM to find data on the internet and populate each value in the new column.

Which of the following platforms do data professionals often use to make interactive dashboards and visualizations? Select the best answer.

• GitHub

• Medium

• Kaggle

• Tableau

A data analyst is reviewing sensor readings from a weather station. The data is represented as a table of rows and columns, where each row represents a date/time and each column is a type of reading (e.g., barometric pressure, temperature, humidity, etc.). Each reading is rendered in a cell that the analyst can click into and manipulate. What type of tool is the analyst using? Select the best answer.

• A database

• A dashboard

• A programming language

• A spreadsheet

Which of the following tools are examples of data visualization tools? Select all that apply.

• Oracle

• Seaborn

• Tableau

• Matplotlib

A financial analyst at an investment firm is monitoring the performance of various investment portfolios. They are using a screen that displays real-time stock prices, portfolio values, risk metrics, and recent news headlines relevant to the holdings. The analyst can access individual securities and customize the display to show different time horizons and performance indicators. What type of tool is the financial analyst most likely using? Select the best answer.

• A programming language

• A dashboard

• A database

• A spreadsheet

Which of the following are limitations of large language models (LLMs) like Gemini or ChatGPT? Select all that apply.

• The information they offer is not always factual.

• They take a very long time to generate content.

• They cannot be relied upon to always make ethical recommendations.

• They are not good at dealing with ambiguity.

When writing a prompt for an LLM, how does it help to try multiple prompts?

• Trying different prompts can provide different perspectives on a problem and enable the LLM to generate a variety of useful responses.

• It allows the LLM to access a wider range of its training data with each new prompt.

• It guarantees that at least one of the responses is completely accurate.

• Each new prompt forces the LLM to use significantly less computational power, making the process more efficient for future queries.