Claude VS ChatGPT For Decision Support Use Cases
The AI hype must translate into high-value applications. Most startups solve big problems really well but flounder when it comes to solving the little problems. Big problems make you famous. Little problems make you rich. Startups need both.
Big problems that make startups famous attract customers, evangelists, promoters, talent, and investment. Those are critical for momentum, but the reason so many AI startups, even in the tools space, have floundered is the applications side of the equation.
Businesses of all sizes need a path to monetize technology, and that’s driving demand for Data Product Managers and Technical Strategists. The new unicorns are people who can translate technology into revenue. Without those people in place, businesses miss opportunities and threats like the ones I’m discussing in this article.
Minor problems are what scale monetization. Only a few companies and individuals need a cutting-edge model's full potential. To scale, startups must leverage the model to provide solutions to much smaller problems.
Decision support applications are a low-glamour, high-value segment for models like Claude and GPT. They are an extension of search and a long overdue improvement. In one decision support use case category, the user is looking for information they don't have. Decision support resolves two access barriers to data.
Time. Background research takes hours or more for simple decisions like the ones I will use next.
Complexity. Current search tools don't support highlighting data categories that improve decisions.
Insights are required to support decisions. Too much of what search provides today are obvious bits of information we already have. The purpose of running a search for decision support is to discover something new.
Defining The Boundary Between Obvious Data And Decision-Supporting Insights
There's a vast difference between providing data and providing insights. Insights help us make a better decision than we would make without them. Data on its own doesn't necessarily produce a better decision outcome. Information overload and analysis paralysis result from having all the data possible but no context or roadmap to help us deploy it to improve our outcomes.
That's the critical difference between data, analytics, and models. Data and analytics are both descriptive. Analytical or descriptive models are capable of surfacing more complex patterns in datasets. This is where we begin to crossover into insight territory. The distinction between an insight and data is the insight's ability to improve the decision outcome.
Big decisions can be broken down into decision chains. No complex decision is made in isolation, and multiple factors (data points to evaluate) and steps (smaller decisions that support the actionable decision) exist. Search and decision support tools are valuable when they target decision chains and provide insights to match. Analytical and machine learning model-driven insights can reveal factors we were unaware of.
Data isn't completely useless. Having information allows us to make better decisions. However, we need to be aware of the relevance of a particular data point for it to lead to an improvement. Analytics and models can do this for us. We need them to point out the most important data points for each link in the decision chain when faced with an open field of data.
Retail Use Cases For Decision Support To Improve Sales
I'm looking for ideas for a Valentine's Day gift in this first example. My wife has a significant amount of Disney merchandise available for purchase. However, there is still an avalanche of products that I could consider (complexity). It would be time-consuming for me to sift through catalogs and websites looking for the perfect gift.
What am I looking for the decision support tool to do for me? In a search, I'm asking Google or Bing to go out and search its knowledge graph for something with specific attributes corresponding to my needs that I've never been shown before. I needed to narrow down the categories and products to interact with the available options more efficiently.
Search tools like Google fall flat in these use cases. I can't put enough parameters into the search engine for it to understand my exact needs. I don't know all the potential parameters that could be involved in making the decision. I need something smarter, and tools like Claude or ChatGPT offer options that could help meet some of these needs.
"My wife is a huge Disney fan but has many of the common gifts from them. What can I get her for Valentine's Day?"
ChatGPT: "You could consider some unique and rare Disney merchandise, such as limited edition collectibles, vintage items, or original artwork. You can also plan a surprise trip to Disneyland or Disney World, or even to a Disney-themed cruise. Another option could be to plan a romantic dinner at a Disney-themed restaurant or a special event, such as the Disney Princess Half Marathon. You could also consider personalized gifts, such as a custom portrait of your wife as a Disney princess or a scrapbook of your Disney memories together."
The last line has insights. I didn't realize getting a custom portrait as a Disney Princess was an option. The scrapbook is another sentimental item that has quite a bit of potential. Both are great suggestions.
In this use case, ChatGPT provides me with a different way of thinking based on the product dataset. It has surfaced insights. Here's the same prompt with Claude:
"Here are some unique Disney gift ideas for a fan who already has many of the common items:
Custom Disney jewelry featuring her favorite character or characters. You can find jewelers online who offer custom Disney-themed necklaces, bracelets, rings, and more.
A Disney art print or framed piece of concept art from her favorite movie. Many artists sell officially licensed Disney artwork.
Disney kitchenware like a Mickey waffle maker or Minnie Mouse kettle. Quirky kitchen appliances and accessories featuring Disney characters can make for unique and fun gifts.
Disney collectibles like limited edition figurines or plush toys. If she has a favorite character, see if you can find a special collectible item from companies like Jim Shore or Precious Moments.
Disney theme park tickets or vacation package. If it's in your budget, a trip to a Disney theme park or resort is a magical experience for any Disney fan. Even just a single day ticket could be a special treat.
A Disney movie or TV show on Blu-ray that she doesn't already own. If she has a favorite Disney film or show, look for a special anniversary or collector's edition that she would enjoy.
Disney-scented products like lotions, perfumes, candles, or reed diffusers. Many companies offer lovely scented products using fragrances inspired by Disney movies and characters."
I hadn't considered the idea of kitchen and bakeware. My wife is into baking, and this is a great idea. Art surfaces again, and I think that's a great avenue for exploration. Customized jewelry is another excellent option.
I hadn't considered all of these. Working with Claude and ChatGPT surfaces insights that help me make a better decision about a Valentine's Day present. Mission accomplished, but my next question will show a clear winner between the two.
More Complex Retail Product Decision Chains
In the last use case, I wanted novel categories for me to explore further. The responses I received helped me to know what to search for. I am still moving to a Google search as the next step in my workflow. These use cases are opportunities for traditional search engines to improve functionality.
In more complex use cases, I need context and information. This time I am deciding about buying a TV. With my wife's present, I have all the decision context necessary to make a great decision. I lack a complete picture of my available options. Claude and ChatGPT handled this case well.
With the TV, I don't have the decision context. All I have are a good description of my current needs. Will either agent support this decision use case?
"I am buying a new 65" TV in the next 2 months. I play a lot of video games. I also watch series like The Last of Us and The Curse of Oak Island. Baking shows are a favorite along with documentaries. What are the most important features to consider so I get the best one for my needs?"
ChatGPT: "When shopping for a 65" TV, there are several features to consider to ensure you get the best one for your needs:
HDR compatibility: If you watch shows and play video games with high dynamic range (HDR) content, look for a TV that is HDR compatible. HDR technology enhances the contrast, color, and brightness of the picture.
Low input lag: For gamers, low input lag is critical to ensure a smooth and responsive gaming experience. Look for a TV with a low input lag of less than 20ms.
Motion handling: If you watch fast-paced content such as sports or action movies, a TV with good motion handling can help reduce motion blur and keep the image clear.
Picture quality: Pay attention to the TV's contrast ratio, color accuracy, and brightness levels. A TV with a high contrast ratio and accurate colors will produce more lifelike images.
Smart TV capabilities: If you want to stream content from popular services such as Netflix, Hulu, or Disney+, look for a TV with built-in smart TV capabilities or the ability to connect to a streaming device.
Audio quality: If you don't have a separate sound system, look for a TV with good built-in speakers. This will enhance your overall viewing experience, especially for shows that have a lot of dialogue or sound effects."
Claude: "Based on your usage, here are the key features to consider when buying a 65-inch TV:
Low input lag and fast response time. For gaming, you want a TV with minimal input lag to avoid controls feeling sluggish. Look for a "game mode" with fast response time.
High contrast and dark blacks. For watching shows and movies with dark scenes like The Last of Us, a high contrast ratio and deep black levels are important for good picture quality. OLED TVs are best for this but more expensive. Some LED-LCD TVs with full array local dimming also handle dark scenes well.
Wide viewing angles. If you tend to watch with a group or from the side at times, wider viewing angles are important so the image remains clear. OLEDs and high-end LED TVs typically have the widest angles.
Accurate and adjustable colors. For watching baking shows and other vivid content as well as playing games, color accuracy and adjustable temperature/color settings are useful. Some TVs have multiple presets while others allow for granular adjustments to fine-tune the image.
Smart TV platform with apps. For streaming shows and documentaries, a smart platform with apps like Netflix, Hulu, and YouTube built-in is very handy. Android TV, Roku TV, and Tizen are all good options.
Multiple HDMI ports. For connecting game consoles, streaming devices, and other components, multiple HDMI ports are needed. At least 3-4 HDMI ports is good for a 65-inch TV."
The best response reveals these models’ true revenue generating potential.