Now that I’ve been using GenAI for over a year, I have acquired some fluency in directing an AI to help me get things done. Yet, AI platforms have upgraded and shifted to become different every few months. For this reason, I also put aside time to test and compare the dominant platforms so I’m not using something obsolete or missing out on critical new capabilities that could save me time.

So, reporting back on which AIs are my preferred platforms and what I use each of them for might be helpful to other nonprofit leaders starting to use AI.

ChatGPT-4(subscription model)  was my most-used AI platform for the longest time, with Claude and Google Bard (now Gemini) as second and third choices. Now that all the platforms have added new skills and more memory, my preferences have shifted. 

Claude Opus(subscription model) is number one for overall writing, analysis, and editing. The language handling and nuance of this latest Claude can continue to surprise and delight me. I also enjoy how human Claude Opus sounds, though it is evident to me that it is a machine trained to reflect and impersonate human characteristics, nothing more. I find myself using it for work and asking it for thoughts about human dynamics and behaviors–with good results. For example, I asked Claude for advice about my disappointment with a friend–and the suggestions were solid.

While Claude excels at general writing tasks, I turn to ChatGPT-4 for more specialized applications. While Claude can outshine ChatGPT-4 in many ways, it doesn’t offer custom GPTs like ChatGPT-4. I use my custom GPTs constantly, especially as an analytical and summarizing tool for large files related to a specific grant or nonprofit organization. My custom GPTs can store data in their uploads and then function in a context where they excel at extracting and analyzing the information I have shared. These capabilities are superb for understanding a new grant RFP or summarizing principles, policies, and outcomes from a nonprofit client. Because my custom GPT can restrict the AI’s scope, my custom GPT can analyze the limited data set I upload without checking the other sources available. This way of using AI will reduce–but does not eliminate–the hallucinations that sometimes appear as made-up information in my results. Until Anthropic delivers this capacity for Claude, ChaptGPT-4 will remain an essential model for me.

ChatGPT-4 is also my preferred illustration maker and visual reference tool. This is as much out of lazy familiarity as anything. I have always been a bit unhappy with what DALL-e makes for me, and I am still trying to find better alternatives.

The new Google Gemini is said to have many strengths, including a more extended short-term memory than the others. My experience with the free version could have been better. I used it for fact-checking some data in a grant proposal with limited success (it told me to check the websites I was trying to avoid using myself to get accurate information rather than pull that information for me). I also proposed that it help me find web-based data, which Bard used to do but failed to do. If I put more time in (and subscribe), I’ll better understand what it can do for me than the others.

What Does My Experience Suggest for You?

If you haven’t used AI  much, select one of these platforms and start experimenting. A newer platform called Poe allows you to try all these different AIs for free. Get started by asking the AI non-work questions, such as help planning a trip, designing a meal plan for a diet, or advising about toddlers and types of strollers (all real-life examples for me). 

Once you have some basic familiarity, consider spending $20 monthly to subscribe to a more powerful, fuller-featured version. If you want to build your custom GPTs or use others’ GPTs, that means OpenAI GPT-4; if you want overall AI work, consider Claude Opus (or whatever else you like).

Resources

Also, a growing part of my consulting practice is coaching nonprofit leaders in using AI, building AI tools, and bringing AI into your organization; if this interests you, get in touch. (Booking time through Calendly is an excellent place to start.)

 

Now that I’ve been using GenAI for over a year, I have acquired some fluency in directing an AI to help me get things done. Yet, AI platforms have upgraded and shifted to become different every few months. For this reason, I also put aside time to test and compare the dominant platforms so I’m not using something obsolete or missing out on critical new capabilities that could save me time.

So, reporting back on which AIs are my preferred platforms and what I use each of them for might be helpful to other nonprofit leaders starting to use AI.

ChatGPT-4(subscription model)  was my most-used AI platform for the longest time, with Claude and Google Bard (now Gemini) as second and third choices. Now that all the platforms have added new skills and more memory, my preferences have shifted. 

Claude Opus(subscription model) is number one for overall writing, analysis, and editing. The language handling and nuance of this latest Claude can continue to surprise and delight me. I also enjoy how human Claude Opus sounds, though it is evident to me that it is a machine trained to reflect and impersonate human characteristics, nothing more. I find myself using it for work and asking it for thoughts about human dynamics and behaviors–with good results. For example, I asked Claude for advice about my disappointment with a friend–and the suggestions were solid.

While Claude excels at general writing tasks, I turn to ChatGPT-4 for more specialized applications. While Claude can outshine ChatGPT-4 in many ways, it doesn’t offer custom GPTs like ChatGPT-4. I use my custom GPTs constantly, especially as an analytical and summarizing tool for large files related to a specific grant or nonprofit organization. My custom GPTs can store data in their uploads and then function in a context where they excel at extracting and analyzing the information I have shared. These capabilities are superb for understanding a new grant RFP or summarizing principles, policies, and outcomes from a nonprofit client. Because my custom GPT can restrict the AI’s scope, my custom GPT can analyze the limited data set I upload without checking the other sources available. This way of using AI will reduce–but does not eliminate–the hallucinations that sometimes appear as made-up information in my results. Until Anthropic delivers this capacity for Claude, ChaptGPT-4 will remain an essential model for me.

ChatGPT-4 is also my preferred illustration maker and visual reference tool. This is as much out of lazy familiarity as anything. I have always been a bit unhappy with what DALL-e makes for me, and I am still trying to find better alternatives.

The new Google Gemini is said to have many strengths, including a more extended short-term memory than the others. My experience with the free version could have been better. I used it for fact-checking some data in a grant proposal with limited success (it told me to check the websites I was trying to avoid using myself to get accurate information rather than pull that information for me). I also proposed that it help me find web-based data, which Bard used to do but failed to do. If I put more time in (and subscribe), I’ll better understand what it can do for me than the others.

What Does My Experience Suggest for You?

If you haven’t used AI  much, select one of these platforms and start experimenting. A newer platform called Poe allows you to try all these different AIs for free. Get started by asking the AI non-work questions, such as help planning a trip, designing a meal plan for a diet, or advising about toddlers and types of strollers (all real-life examples for me). 

Once you have some basic familiarity, consider spending $20 monthly to subscribe to a more powerful, fuller-featured version. If you want to build your custom GPTs or use others’ GPTs, that means OpenAI GPT-4; if you want overall AI work, consider Claude Opus (or whatever else you like).

Resources

Also, a growing part of my consulting practice is coaching nonprofit leaders in using AI, building AI tools, and bringing AI into your organization; if this interests you, get in touch. (Booking time through Calendly is an excellent place to start.)