Two AI Assistants, One Question — What Can We Learn from Their Different Perspectives?
In Episode 1, we explored an interesting idea: using ChatGPT and Claude together, with ourselves acting as the human director.
Today, let's take the idea one step further.
Instead of asking one AI assistant a question and accepting its answer, why not ask two AI assistants exactly the same question?
Then compare their answers.
Would they give us the same recommendations? Would one offer ideas the other overlooked? And, more importantly, could we learn something useful from the differences?
I decided to try this experiment using a subject very close to my heart: AI123.ca, my blog about AI learning, educational technology and AI productivity, with a special interest in helping seniors.
Let's see what happened.
1. The experiment: Ask the same question
I gave both ChatGPT and Claude exactly the same instructions.
Here was my question:
"I operate a blog called AI123.ca, focusing on educational technology, AI learning and AI productivity, with a special interest in helping seniors.
Analyze opportunities to improve its content, readership and potential affiliate income.
Give me five practical recommendations, ranked by importance, explaining why each recommendation matters."
Nothing complicated. No special instructions favouring one assistant over the other.
I simply wanted to see how two AI assistants would approach the same business and publishing challenge.
2. What ChatGPT suggested
ChatGPT approached the question from a broader publishing and business development perspective.
It identified five priorities:
1. Build a recognizable content identity.
It suggested organizing AI123.ca around four main areas:
AI Learning for Seniors
AI Productivity
AI in Everyday Life
AI123 Digest
I particularly liked this recommendation because it supports the direction I have been developing for the blog.
2. Turn visitors into regular readers.
Getting people to visit a website is one thing. Giving them a reason to return is another.
ChatGPT suggested improving internal links, connecting related articles and eventually developing a regular AI123 Digest newsletter.
3. Develop affiliate income through useful recommendations.
Rather than joining every available affiliate program, it recommended concentrating on products that genuinely help readers.
For example, demonstrating how Pictory can turn a written article into a video is more convincing than simply advertising video software.
4. Improve Google search visibility.
ChatGPT recommended creating practical, evergreen articles that answer questions people are already asking.
Examples include using ChatGPT for beginners, AI writing assistance and learning to work with multiple AI assistants.
5. Build trust before expanding revenue opportunities.
This was another recommendation I appreciated.
ChatGPT suggested developing a beginner's guide, an AI tools directory and a recognizable publication.
Its overall message was simple: build something readers find useful and trustworthy before concentrating too heavily on making money.
3. What Claude suggested
Claude approached the same question somewhat differently.
Its recommendations were more focused on the reader's experience and the practical details of running a website.
Here were its five priorities.
1. Make articles easier to find through Google.
Claude suggested using titles that reflect what people actually search for.
For example, instead of a general title such as "Illustration 2", a more useful title might be:
"How to Write Your First ChatGPT Question: A Step-by-Step Guide for Seniors."
It also suggested developing more articles about online safety, voice cloning, suspicious emails and checking AI-generated information.
2. Build an email subscriber list.
Claude noticed that AI123.ca could benefit from a more convenient way for visitors to keep in touch.
It suggested using the free ebook as an incentive for people to subscribe to a regular email newsletter.
I thought this was worth considering.
People may discover an article through Google, but an email newsletter gives us another opportunity to maintain a relationship with them.
3. Make affiliate recommendations more relevant.
Claude raised an interesting point.
Affiliate products should match the interests and needs of the readers.
Someone learning to use ChatGPT may not necessarily be interested in an expensive computer docking station.
Instead, practical products such as keyboards, tablets, password managers and useful AI software might be more relevant.
It also suggested exploring local advertising opportunities with businesses serving seniors in Markham and Toronto.
4. Improve the mobile reading experience.
This was a particularly practical observation.
Many seniors use smartphones and tablets to read online.
Claude suggested checking the website's mobile appearance, improving text readability, increasing contrast and reducing unnecessary sidebar distractions.
It reminded me that a website should be comfortable to read, not simply attractive on a desktop computer.
5. Reach seniors through their existing communities.
This was perhaps the most interesting difference between the two assistants.
Claude suggested taking AI learning beyond the internet.
Libraries, seniors' centres, community groups and social gatherings could become useful places to introduce AI123.ca.
I remembered our recent lunch gathering where we discussed artificial intelligence. People were curious about ChatGPT, self-driving cars and how AI could help with everyday writing.
Perhaps there are opportunities to share what I have learned with more people in similar situations.
4. Comparing the two answers: What did I learn?
After reading both responses, I noticed something important.
ChatGPT and Claude were not necessarily disagreeing. They were looking at the same challenge from different angles.
ChatGPT concentrated more on the overall direction of the publication, its content structure and long-term development.
Claude concentrated more on practical improvements, website usability, attracting readers and reaching people outside the internet.
Here is my comparison:
| Area | ChatGPT | Claude |
|---|---|---|
| Content strategy | Strong focus | Search-focused |
| Readership | Returning readers | Email and community |
| Affiliate income | Content-based recommendations | Audience relevance |
| Website development | Long-term publication | Mobile experience |
| Growth opportunities | Brand and educational resources | Local outreach |
Which answer was better?
I would not say one was necessarily better than the other.
They were useful in different ways.
ChatGPT helped me look at the bigger picture. Claude encouraged me to examine some practical details that I might otherwise overlook.
That is precisely why I find this experiment worthwhile.
5. The human director still makes the decisions
There is one important lesson I want to emphasize.
Having two AI assistants does not mean we should automatically follow twice as many recommendations.
Some suggestions may be useful immediately. Others may be worth considering later. A few may not fit our circumstances at all.
For example, I agree with building a stronger content identity, improving the reading experience and making affiliate recommendations more relevant.
However, I don't want to rush into launching a newsletter, expanding advertising or creating too many new projects at once.
AI123.ca already has a direction:
Educational technology and AI productivity, while retaining its special focus on helping seniors learn AI.
I prefer to keep that direction and add value gradually.
After all, running a blog should be an enjoyable learning experience, not another source of pressure!
This is where human judgment becomes important.
AI can generate ideas, compare possibilities and point out things we might have missed. But we still decide what matters, what is practical and what deserves our time.
6. Try this experiment yourself
You don't need to operate a blog to enjoy this exercise.
Choose something you are interested in, perhaps improving your daily routine, planning a holiday, learning a new skill or organizing your household.
Then try this simple exercise.
Step 1: Open ChatGPT and Claude.
Step 2: Give both assistants exactly the same question.
Step 3: Read their answers carefully.
Step 4: Compare their recommendations.
Ask yourself:
What did both assistants agree on?
What did one suggest that the other overlooked?
Which suggestions are practical for my situation?
What would I like to try first?
You don't need to accept everything either assistant tells you.
The purpose is to get different perspectives and make a better-informed decision.
My final thoughts
I started this experiment simply because I was curious about using two AI assistants together.
What I discovered was more interesting than I expected.
ChatGPT and Claude can complement each other. One may offer a broader perspective, while the other may identify practical details worth examining.
And we don't have to be technology experts to benefit from this approach.
At 80+, I am still learning, experimenting and discovering new ways to use artificial intelligence.
I find this particularly encouraging. Learning AI does not have to be about understanding complicated technology. Sometimes, it is simply about asking better questions and being willing to explore different answers.
Perhaps that is the real meaning of AI productivity.
Don't just ask AI to give you an answer. Ask it to help you think, compare and make better decisions.
And remember, you are still the director. AI is your assistant.
Let's continue exploring!
AI123.ca — Exploring → Learning → Documenting