The best way to learn AI is to, in the words of Professor and AI Expert Ethan Mollick, “invite it to the table.” The more you use AI for personal and professional tasks, the faster you’ll see the power and potential of the AI revolution. Whether we like it or not, AI is changing the way we work and live, so gaining some familiarity with AI will be critically important for most of us.
Animal shelters are already using AI to:
- Augment customer service to ensure consistency and accuracy
- Develop marketing and communications plans and campaigns
- Write and update policies, procedures, and other organizational documents
- Analyze and better use your shelter data
- Streamline the flow of animals through the system to decrease length of stay
- Get more lost pets home more quickly
- Edit images without changing the appearance of the animal itself
- Effectively manage risk and improve shelter safety
If you’re just getting started or you’re training others, here are some simple tips to make learning AI fun and easy. To accelerate your learning, sign up for our five-week AI For Animal Welfare Professionals course.
Start by using at least two large language models (LLMs). These include Anthropic’s Claude, Google Gemini, and Chat GPT. You can also use a platform like Microsoft Copilot which is powered by these other LLMs and is a commonly-approved platform for governments and nonprofits. It’s a good idea to use at least two different platforms so you can see the different results you get with each one. Use the paid versions, which currently average about $20 per month, while you’re getting started because some of the best features are only available to paid users. Check out my list of platforms and resources for all kinds of tasks.
Provide context. Avoid the beginner’s mistake of simply uploading a document and asking AI to “make it better.” If you want to design a web page that looks just like the one you saw the other day, provide that web page to AI and tell it to make yours look similar. If you want AI to write a board report, share an example of an earlier report, tell AI who your audience is, and what you really want them to remember. When we ask AI to help us create, update, and edit work products, the more specifics we can give, the better the end result will be.
Use the right tool or platform. Some LLMs excel at editing images. Others are great at writing and coding. Still others are the best choice if you want to analyze data. At the time I’m writing this, Claude excels at written documents but Google Gemini is better for editing images. When starting a new project, make sure you’re using the best tool for the job. You can find this out by simply Googling or asking Perplexity, an AI-powered search engine. You can say something like, “What is currently the best AI tool for creating videos?” You’re likely to get some advertisements, but you’ll also get a list of platforms that can do the best job with the particular thing you’re asking.
Know what you want the end result to look like. This is a tough one to explain but you’ll know when it happens to you. You ask AI to do something. You get five to 10 pages of information back, and it’s all technically correct but it’s not what you wanted. Stop and ask yourself, “What did I want when I prompted AI?” There is a good chance you either did not provide enough context, you weren’t sure what you wanted, or you asked the wrong question. If this happens, go back to the beginning. You might even consider starting a new chat so you can clean your mental slate. AI is good enough now that it can get to the right answer, but you might need to get clearer on what you want to end up with.
Not sure what to do? Ask AI. When you inevitably get ‘stuck’ while using an LLM or another AI tool, ask AI for help. For example, if I’m working on analyzing data in a platform like Claude or Julius, I can say to AI, “How do I figure out what data points will show the most meaningful change in intakes and outcomes?” At that point, the platform will probably provide you with a list of data points it can produce and you can choose the ones that you think will be the most helpful.
Check your work. AI sometimes hallucinates, which means it will not always give you the right answer. Two things you should always check: First, make sure the information the LLM is producing is based on research and data. You can do this by asking the LLM to do research before responding. Even better, you can tell it the specific sources you want it to research. I was recently helping a student write a cat and kitten foster manual. He had created the manual using AI but had not told it the specific sources to research. When we told AI to use the Association of Shelter Veterinarians Guidelines and the Maddie’s Million Pet Challenge, we got a better foster manual based on the latest guidance from the experts. Second, always check the citations. AI will make up citations even when real ones don’t exist so make sure the source it’s referencing really exists!
Learn basic AI concepts. Prompting is how we ask AI to do things. LLMs are the foundational models upon which AI-powered platforms are built. A chatbot uses AI models to conduct conversations in text and audio. There is an ever-expanding vocabulary of AI terms, and terminology changes over time. The best way to familiarize yourself with AI terminology is to listen to an AI podcast like the AI Daily Brief or How I AI, or Hardfork, which is a tech podcast that focuses heavily on how AI is changing the tech landscape. The more you learn about how AI works, the more you’ll be able to deal with stuck points and challenges. AI proficiency makes AI less frustrating and more fun!
Balance benefit and risk. Many AI platforms integrate seamlessly into your email, calendars, and cloud-based or personal drives. This type of integration can increase your risks when it comes to sensitive information. Avoid sharing social security numbers, passwords, human resources files and other personal information with LLMs. Some organizations are developing AI-use policies which may include specific guidance on what platforms are permitted, what training is required, and what information can and cannot be shared with AI. While organizations should be aware of and mitigate risks associated with AI, it’s also important to recognize that simply banning or severely restricting AI use will have increasingly negative consequences for animal welfare organizations. The greatest protection we have is understanding how AI works and using it appropriately and safely.
Kristen