Categories
AI AI: Large Language Models Bing Chat ChatGPT Claude Google Bard

How Much Pepperoni?

pizza on brown wooden board
Photo by Brett Jordan on Pexels.com

I was intrigued by this sentence in a recent column by Bloomberg’s Matt Levine: The stereotypical way that a management consulting interview works is that you get asked questions like โ€œhow much pepperoni  is sold in the US every year,โ€ and then you reason your way to some estimate of the answer.

When I read this, I thought it would be fun to play with my favorite chatbots again and see if they might help me reason my way to an answer. I got very different results – not surprisingly. The prompt I used was:

The stereotypical way that a management consulting interview works is that you get asked questions like โ€œhow much pepperoni  is sold in the US every year,โ€ and then you reason your way to some estimate of the answer. Let’s reason step by step to try to find an answer to this question.

First, here’s what I got back from Google Bard:

Sure, here is a step-by-step process of how to reason our way to an estimate of how much pepperoni is sold in the US every year:

  1. Gather some basic facts. We know that the average American eats 24 pounds of pizza per year, and that pepperoni is the most popular pizza topping, accounting for about 30% of all pizza orders. We also know that there are about 330 million people in the US.
  2. Make some assumptions. We need to make some assumptions about how much pepperoni is used per pizza. A conservative estimate might be that each pizza uses 1 ounce of pepperoni. This means that the average American eats about 2.4 ounces of pepperoni per year.
  3. Do some math. We can now estimate that the total amount of pepperoni sold in the US every year is about 800 million ounces. This is equivalent to about 250 million pounds.

This is just an estimate, of course, but it is based on some reasonable assumptions. It is likely that the actual amount of pepperoni sold in the US is higher than this, but it is also possible that it is lower. More accurate estimates could be made by collecting more data and making more refined assumptions.

Here are some additional factors that could be considered to improve the accuracy of the estimate:

  • The amount of pepperoni used per pizza can vary depending on the size of the pizza, the type of pizza, and the restaurant.
  • The popularity of pepperoni may vary from region to region in the US.
  • The amount of pepperoni imported into the US may also be a factor.

By considering these factors, it is possible to make a more accurate estimate of the amount of pepperoni sold in the US every year.


Next up was ChatGPT:

Of course, I’d be glad to guide you through the process of reasoning to estimate the amount of pepperoni sold in the US every year. It’s important to note that management consulting interviews often focus on assessing your problem-solving skills and ability to think logically. Let’s break down the problem step by step:

Step 1: Clarify the Scope Before we begin, it’s essential to clarify the scope of the question. Are we looking for the total weight of pepperoni sold, or the total revenue generated from its sales? This distinction will influence the data sources we’ll need to consider.

Step 2: Identify Relevant Data Sources To estimate the amount of pepperoni sold in the US, we’ll need to gather data from reliable sources. These could include industry reports, government statistics, market research firms, and food industry associations. Some possible sources might be the U.S. Department of Agriculture (USDA) and organizations like the American Meat Institute.

Step 3: Estimate Total Consumption One approach could be to start with the total meat consumption in the US and then estimate what percentage of that is likely to be pepperoni. You might need to consider factors such as the popularity of pizza, sandwiches, and other dishes that use pepperoni. Additionally, think about variations in consumption due to cultural and regional preferences.

Step 4: Break Down Consumption Break down the total estimated consumption of pepperoni into categories such as retail sales (supermarkets and grocery stores) and food service (restaurants, pizzerias, etc.). This can help refine your estimate by considering different distribution channels.

Step 5: Estimate Quantity per Serving and Servings Estimate the average quantity of pepperoni used per serving. For example, how many slices of pepperoni are typically used on a pizza or in a sandwich? Then, multiply this by the estimated number of servings sold in a year.

Step 6: Adjust for Factors Consider any factors that might affect the estimate, such as seasonal variations in consumption (for instance, more pepperoni might be consumed during holidays or certain events) or any recent trends that might have impacted consumption patterns.

Step 7: Calculate the Estimate With all these factors in mind, you can proceed to calculate your estimate. Make sure to show your calculations and assumptions clearly.

Step 8: Validate and Compare After arriving at your estimate, it’s a good practice to validate it against any available data, if possible. Compare your estimate to industry reports or official statistics to assess its accuracy.

Remember, the goal of this exercise is not necessarily to arrive at an exact figure, but to showcase your ability to logically reason through a complex problem using available information. Feel free to ask questions if you need further clarification on any of these steps.


Next up is Claude:

Okay, let’s break this down:

First, we need to estimate how much pizza is sold in the US each year. Pizza is a very popular food, so let’s assume that on average, each American eats pizza once per week. With a population of around 330 million, that’s about 330 million pizzas sold per week, or around 17 billion pizzas per year.

Next, we need to estimate what percentage of those pizzas have pepperoni. Pepperoni is a very common pizza topping, so let’s assume it’s on 50% of all pizzas sold. That brings us to around 8.5 billion pepperoni pizzas per year.

Now we need to estimate how much pepperoni is on each pizza. Let’s assume an average pepperoni pizza has around 50 pepperoni slices. If each slice is roughly 1 ounce, that’s about 50 ounces or around 3 pounds of pepperoni per pizza.

With 8.5 billion pepperoni pizzas per year, each with around 3 pounds of pepperoni, that gives us an estimate of around 25.5 billion pounds of pepperoni sold annually in the US.

Of course this is just a rough estimate with a lot of assumptions. The actual amount may vary quite a bit. But walking through the logical steps and making reasonable assumptions is generally the approach for these types of brainteaser questions. Let me know if this helps explain the thought process!


Finally is Bing Chat:

Thatโ€™s an interesting question. To estimate how much pepperoni is sold in the US every year, we can use some assumptions and data from the web search results. Here are some possible steps to follow:

  • First, we can assume that most of the pepperoni sold in the US is used as a pizza topping, since it is one of the most popular pizza toppings.
  • Second, we can use the data from 7 to estimate how much pepperoni is used on a pizza of different sizes. For example, a 12-inch pizza uses about 1.75 ounces of pepperoni, and a 18-inch pizza uses about 3 ounces of pepperoni.
  • Third, we can use the data from 6 to estimate how many pizzas are sold in the US each year. According to this source, there are 3 billion pizzas sold in the US each year.
  • Fourth, we can use the data from 13 to estimate what percentage of pizzas ordered are pepperoni. According to this source, 36% of people order pepperoni on their pizzas.
  • Fifth, we can multiply the number of pizzas sold by the percentage of pepperoni pizzas, and then multiply by the average amount of pepperoni per pizza. To simplify the calculation, we can assume that the average pizza size is 14 inches, which uses about 2.5 ounces of pepperoni according to 7.
  • Therefore, the estimated amount of pepperoni sold in the US each year is:

3 billion x 0.36 x 2.5 = 2.7 billion ounces

  • To convert this to pounds, we can divide by 16, since there are 16 ounces in a pound. This gives us:

2.7 billion / 16 = 168.75 million pounds

  • This is our final answer, based on the assumptions and data we used. Of course, this is only an approximation, and there may be other factors that affect the actual amount of pepperoni sold in the US each year.

I hope this helps you understand how to approach this kind of question. Do you have any other questions for me? ๐Ÿ˜Š


Bard estimated 250 million pounds. Claude estimated 25.5 million pounds. Bing estimated 168.75 million pounds. ChatGPT didn’t provide an estimate – it just discussed a methodology to get to an answer (which is what’s more important in a job interview!).

What’s the correct answer? Some quick Google searching finds results around 250 million pounds a year – closest to Bard’s answer.

What can we learn from this little exercise? I would have expected both Bard and Bing Chat to have an advantage as they’re both capable of doing web searches while ChatGPT (3.5) and Claude aren’t. Bing actually cites specific web references. But the most useful result actually isn’t the final value – all of the chatbots demonstrated steps to go through to get to an answer and it’s the ability to demonstrate those steps that answers the interview question. I’m sure by continuing each of the conversations I could have gotten more insights from each of the chatbots. Meanwhile, I’m now hungry for a pepperoni pizza!


Update 04/21/2024: a few days ago Meta released its new Llama 3 model so I tried my original prompt out on the Meta.ai website. Hereโ€™s what I got back:

The classic “pepperoni problem”! Let’s break it down step by step. Here’s a suggested approach:

Step 1: Define the scope
Clarify what’s meant by “pepperoni sold in the US every year.” Are we including:

  • Only pepperoni slices sold as a standalone product?
  • Pepperoni used as an ingredient in other products (e.g., pizzas, sandwiches, salads)?
  • Both domestic consumption and exports?

For this example, let’s assume we’re focusing on domestic consumption of pepperoni as a standalone product or as an ingredient in other food products.

Step 2: Identify key drivers
What factors influence pepperoni sales in the US?

  • Population growth and demographics
  • Pizza consumption habits (as pepperoni is a popular pizza topping)
  • Trends in food preferences and dietary habits
  • Competition from alternative meat products

Step 3: Gather data points
Collect relevant data to inform our estimate:

  • US population: approximately 333 million (as of 2024)
  • Pizza consumption: average American consumes 23 pounds of pizza per year (according to a 2020 survey)
  • Pepperoni’s market share among pizza toppings: assume around 30-40% (as it’s a popular topping)

Step 4: Make assumptions and calculations
Based on the data points, make some assumptions and calculations:

  • Assume an average pepperoni slice weight: 2-3 grams ( rough estimate)
  • Estimate the total number of pepperoni slices consumed per year:
    • 333 million people x 23 pounds of pizza per year รท 2-3 grams per pepperoni slice = approximately 12-15 billion pepperoni slices per year
  • Convert this to a weight-based estimate:
    • 12-15 billion slices x 2.5 grams (avg. slice weight) = 30-37.5 billion grams or 30,000-37,500 metric tons of pepperoni per year

Step 5: Refine and validate
Refine our estimate by considering additional factors, such as:

  • Seasonal fluctuations in pepperoni demand (e.g., summer BBQs, holiday parties)
  • Regional differences in pepperoni consumption
  • Market trends and competition

To validate our estimate, we could research industry reports, market studies, or sales data from major pepperoni manufacturers or distributors.

By following these steps, we’ve arrived at a rough estimate of 30,000-37,500 metric tons of pepperoni sold in the US every year. Keep in mind that this is a simplified example and actual numbers may vary based on more detailed analysis and data.

Categories
AI AI: Large Language Models AI: Prompting Claude

Multi-step prompting for Claude

webpage of chatgpt a prototype ai chatbot is seen on the website of openai on iphone or smartphone
Photo by Sanket Mishra on Pexels.com

I’ve been experimenting with all of that available chatbots including Google’s Bard, OpenAI’s ChatGPT, Microsoft Bing Chat, and Anthropic’s Claude. A few days ago I was playing with Claude when I decided I’d try to learn more about multi-step prompting.

A recent video with Ethan Mollick emphasized that we tend to overemphasize initial prompts – looking for the “perfect prompt” – when what really adds value to a chatbot result isn’t the perfect initial prompt but rather the interaction you have with the chatbot. In other words, just like with an real intern, the conversational back and forth is what really adds to the learning and understanding from a session. I’ve begun applying this approach more regularly as I continue exploring these amazing new tools.

Below is the result of my back and forth conversation with Claude as I tried to learn more about the best practices for multistep prompting.

Introduction: The Power of Multistep Prompting

When conversing with Claude, you can get much more natural, detailed responses by using multistep prompts. This technique involves breaking down complex requests into multiple simpler follow-up prompts, rather than asking lengthy, dense questions all at once.

For example, instead of saying:

“Claude, can you tell me about the major battles of World War 2 fought by American troops in the European theater, with a focus on key events like the Normandy invasion?”

You would ask:

“Claude, what were the major battles American troops fought in during World War 2 in Europe?”

Then based on Claude’s initial response mentioning D-Day and the Battle of the Bulge you could ask focused follow-ups like:

“Can you provide more details on the D-Day invasion?”

“What happened at the Battle of the Bulge and why was it significant?”

This prompts Claude to provide more detailed, robust responses on a specific aspect of your original broad question. Chaining these bite-sized prompts together allows for a natural dialogue where you guide the conversation based on Claude’s replies.

The rest of this post will provide tips to help you get the most out of using multistep prompts to have engaging, productive conversations with Claude. Let’s get started!

Tip 1: Start with a Focused Initial Prompt

When first engaging Claude, begin with a clear, concise prompt setting the context. For example:

“Claude, can you recommend a good sci-fi movie from the last few years that has cool visual effects?”

Starting the conversation with a specific question about the type of movie you want gets better results than a vague prompt like “Let’s talk about movies.” It provides Claude with clear criteria to focus its initial response on recent sci-fi films with great special effects.

Tip 2: Ask One Question at a Time

Deconstruct big requests into individual follow-up prompts. Asking “What were the major events of WW2 and how did technology impact them?” overloads Claude. Instead try:

“What were the major events of WW2?”

“How did technology like radar and rockets impact these events?”

Simpler prompts allow more detailed, thoughtful responses from Claude.

Tip 3: Build on Previous Responses

Review Claude’s last response before asking your next prompt. For example:

You: “Who were Apple’s founders?”

Claude: “Apple was founded in 1976 by Steve Jobs, Steve Wozniak and Ronald Wayne.”

You: “What specific contributions did Steve Wozniak make to Apple?”

This builds a logical flow and conversation.

Tip 4: Clarify When Confused

If Claude’s response doesn’t make sense, politely ask for clarification before moving on:

You: “What was the impact of Midway on WW2?”

Claude: “The Battle of Midway enabled the Allies to gain the upper hand in the Pacific theater.”

You: “I don’t understand how Midway gave the Allies an advantage. Can you explain that more clearly?”

Don’t just accept unclear responses. Ask follow ups to get back on track.

Tip 5: Change Direction Based on Responses

Let Claude’s replies guide where you take the conversation nextโ€ฆ

Tip 6: Periodically Summarize Key Facts

Restating important details helps ensure Claude stays on topic. For example:

You: “Can you tell me about the first airplane flight?”

Claude provides response.

You: “So the Wright brothers flew for the first time in 1903 in Kitty Hawk. What happened next in early aviation?”

Summarizing facts focuses the dialogue and confirms Claude’s understanding.

Tip 7: Keep Conversations 3-5 Prompts Long

Try to complete conversations within 3-5 focused prompts. For example:

Prompt 1: Ask for book recommendations

Prompt 2: Clarify if you want fiction or non-fiction books

Prompt 3: Ask for more details on a specific recommended book

Prompt 4: Thank Claude and end the conversation

While more than 5 exchanges often indicates a disjointed or unclear goal, I often end with a final prompt to the effect of “Is there anything else I should know or try to understand?” In other words, remember perhaps there’s “one more thing!”

Tip 8: Wrap Up with a Summary

End conversations by asking Claude to summarize its overall response:

You: “Can you please summarize the key points you mentioned about World War 2?”

This provides a clean conclusion to the dialogue.

Tip 9: Provide Feedback if Responses Seem Off

If Claude seems confused or makes incorrect statements, politely clarify to improve its training:

You: “I asked about D-Day, but your response seems to be about a different battle. Can you please double check your information about D-Day?”

This constructive input helps Claude improve.

Tip 10: Practice Makes Perfect

Have fun practicing multistep conversations on topics like movies, recipes, sports, current events, trivia, and product recommendations. The more you use prompts and follow-ups, the more natural conversations will become.

Taking It to the Next Level: Advanced Multistep Prompting

While effective for everyday conversations, multistep prompts can also enable advanced, niche uses of Claude you may not have considered:

Foreign Language Translation – Claude has some ability to translate between languages when you provide context. Try prompts like:

“Claude, can you translate this Spanish phrase into English: ‘ยฟCรณmo estรกs?'”

Poetry – Claude can generate original poetry if you guide it. Prompt it to start a poem, suggest a topic, and refine the wording.

Research Assistance – Ask Claude to find and summarize information on obscure topics:

“Claude, can you research the ancient Sumerian civilization and summarize their key achievements?”

Creative Writing – Claude can craft short stories if you provide plot points and characters. Slowly build up a narrative through prompts.

Joke Telling – Give Claude a topic and have it come up with witty, original jokes through multistep exchanges.

Troubleshooting – Walk through a technical problem by describing symptoms and responding to Claude’s follow-up questions.

The possibilities are endless when you get creative with prompts! The key is providing context and steering the conversation to get Claude’s best work.

Conclusion: The Key is Conversation

The core takeaway is that multistep prompting allows for genuine back-and-forth conversation with Claude. Start applying these tips to your prompts and soon you’ll be chatting with Claude like an old friend! I’d love to hear of any other multistep best practices you discover. Now go have some fun conversations with Claude or your other favorite chatbot!