Report on the Use of AI in Animal Advocacy

This report analyzes the findings from a survey that Open Paws conducted amongst animal advocacy organizations to understand their current use of artificial intelligence (AI), the challenges they face, and their attitudes toward integrating AI into their advocacy efforts.

The survey received responses from 194 participants representing 142 organizations, providing a comprehensive view of AI’s potential and barriers in the animal advocacy movement.

Respondent Demographics

The survey attracted responses from individuals in diverse roles:

• Senior leadership - 55.2%

• Staff member - 14.4%

• Middle management - 8.8%

• Volunteer - 5.2%

• Other - 16.4%

This variety ensures that the insights reflect a wide range of perspectives from different organizational levels.

“Our biggest problem is lack of domain experts with time to devote to AI tools. The second biggest is the speciesist biases, factual inaccuracies and logical inconsistencies perpetuated by LLMs.” - Survey Respondent

Current Use of AI in Animal Advocacy

The frequency of AI use among respondents varied:

• Weekly Use: 29.4%

• Daily Use: 19.6%

• Monthly Use: 18.6%

• Very Rarely: 16.5%

• Never: 16%

Despite the significant potential of AI to enhance productivity and effectiveness, nearly half of respondents rarely or never use AI in their advocacy work. This suggests a considerable opportunity for increasing AI adoption through targeted training, tools, and resources.

“I’m sure there are more tasks I do that could be aided by AI but I don’t always know how to use it more effectively.” - Survey Respondent

Attitudes Toward AI

• Personal Attitudes:

  • Positive (7-10): 72.2%

  • Neutral (4-6): 23.2%

  • Negative (0-3): 4.6%

• Organizational Attitudes:

  • Positive (7-10): 59.9%

  • Neutral (4-6): 33.8%

  • Negative (0-3): 6.4%

There is a strong personal and organizational inclination towards AI, with over 70% of individuals and nearly 60% of organizations having a positive outlook.

However, the gap between personal and organizational positivity indicates that while individuals may be open to AI, organizational adoption lags behind.

“I think our main barrier is just trying to adapt to an AI world - to have the question "how could AI help me here?" naturally pop into our heads when confronted with a new problem.” - Survey Respondent

Current Applications of AI

Respondents reported using AI for various tasks, with the most common applications being:

• Writing or Editing Social Media Posts, Blogs, or Press Releases: 50.5%

• Brainstorming and Generating Ideas: 55.2%

• Creating or Editing Images, Audio, or Video Content: 23.7%

• Searching for Reliable Sources of Information: 27.3%

• Personalizing Responses to Comments, Emails, or Messages: 28.4%

• Grant Writing and Fundraising Support: 20.1%

• Automating Routine Tasks and Data Entry: 18.6%

The widespread use of AI for content creation and brainstorming indicates its role in enhancing creative processes. However, more technical applications, such as predicting campaign outcomes, are far less common, suggesting a need for further development and training in these areas.

“I'm lacking the knowledge to use it correctly and efficiently. I think a workshop or course would be very helpful.” - Survey Respondent

“AI often misunderstands my needs. It generates a lot of detailed content that isn't always useful.” - Survey Respondent

Desired Future Uses of AI

Respondents expressed interest in expanding AI applications to include:

• Automating Routine Tasks and Data Entry: 53.1%

• Predicting Campaign Outcomes: 50.5%

• Improving Digital Advertising Performance: 45.4%

• Offering Personalized Support: 31.4%

There is a clear interest in leveraging AI for more sophisticated tasks such as predictive analytics and digital advertising.

This demonstrates the need for specialized and niche-specific AI models trained on campaign and advertising data from animal advocacy organizations, as well as training and education on how to use them effectively.

“My main challenge is not knowing enough about it or how to really use it. For example, you've added a check box in this survey for "Improving digital advertising performance" and I didn't even know AI could help with that!” - Survey Respondent

“It is mostly a lack of technical knowledge (or time to get it) to use AI for more complicated tasks (e.g. brainstorming, redefine strategies or predict campaign outcomes)” - Survey Respondent

Daily, weekly, and monthly users of AI show a significantly higher interest in expanding AI applications compared to those who rarely or never use it.

Specifically, 61.9% of frequent users are interested in predicting campaign outcomes, while only 29.4% of rare users share this interest.

Similarly, 34.1% of frequent users are inclined towards using AI to personalize support, compared to 26.5% of rare users.

Improving digital advertising performance is desired by 50.8% of frequent users versus 35.3% of rare users.

Additionally, automating routine tasks and data entry is favoured by 64.3% of frequent users, significantly higher than the 32.4% of rare users.

These numbers highlight a clear trend where frequent users are more inclined towards utilizing AI for these future tasks compared to those who rarely or never use AI.

Barriers to AI Adoption

The main challenges faced by respondents in using AI include:

• Lack of Technical Expertise: 60.3%

• Hallucinations and False Information in AI Responses: 43.3%

• Limited Budget and Resources: 38.7%

• Data Privacy and Security Concerns: 28.9%

• Speciesist Bias in AI Systems: 21.1%

Daily, weekly, and monthly users of AI indicated all barriers more frequently, such as lack of technical expertise (62.7% vs. 60.3% baseline), hallucinations (51.6% vs. 43.3% baseline), limited budget (44.4% vs. 38.7% baseline), data privacy (31% vs. 28.9% baseline), compliance concerns (15.1% vs. 16% baseline), and speciesist bias (25.4% vs. 21.1% baseline).

“Our biggest concern is hallucinations and lack of compassionate data for AI to utilize…The vast majority of data for farmed animals is biased towards their exploitation, so we manually must use our expertise and cross reference to filter out the bad and keep the good.” - Survey Respondent

“I often have to challenge AI to consider another points of view but it invariably comes back to AI's "insistence" that humans are superior to all other life forms” - Survey Respondent

“Budget is a big one. Ours is currently tight, and has been a limiting factor on hiring external consultants to support with an AI strategy and/or support team members with access to paid platforms.” - Survey Respondent

“(I) feel like AI is part of the system or programmed by people with a total speciesist view of the world, so I do not trust it.” - Survey Respondent

Conversely, those who rarely or never use AI generally show lower awareness across these barriers, with percentages such as 55.9% for technical expertise, 27.9% for hallucinations, 27.9% for budget constraints, 25% for data privacy, 14.7% for compliance, and 13.2% for speciesist bias.

Data Sharing Willingness

82.5% of respondents indicated a willingness to share at least one type of data for training AI models, with the most common being:

• Website Content: 35.1%

• Social Media Content: 32%

• Research and Reports: 20.1%

• Campaign Materials: 14.9%

The willingness to share data for AI training indicates a collaborative spirit within the animal advocacy community. This data could be invaluable in developing AI tools tailored specifically to the needs of animal advocacy, enhancing the overall effectiveness of these tools.

Daily, weekly, and monthly AI users are more willing to share data than those who rarely or never use it.

For private data sharing, 42.1% of frequent users are willing to share website content, 35.7% social media content, 22.2% research and reports, and 15.1% campaign materials, compared to 22.1%, 25%, 16.2%, and 14.7% for rare users.

For public data sharing, 34.1% of frequent users are willing to share website content, 27% social media content, 18.3% research and reports, and 11.9% campaign materials, compared to 19.1%, 19.1%, 14.7%, and 11.8% for rare users.

Individuals with positive attitudes towards AI are also more willing to share data. For example, 39.2% with positive attitudes are willing to share website content privately, and 31.4% publicly, compared to 24.1% and 22.22% for those with neutral or negative attitudes.

This shows that frequent AI use and positive attitudes towards AI correlate with a higher willingness to share data.

Conclusion and Recommendations

The survey reveals a strong interest and positive attitude towards AI among animal advocacy organizations, coupled with significant challenges related to technical expertise, accuracy, and budget constraints. To foster further AI adoption in animal advocacy, we suggest:

1. Providing Technical Training and Education: Implement targeted training and education programs that focus on practical AI applications within animal advocacy.

   1. Workshops, Webinars & Talks: Open Paws has conducted multiple introductory talks and workshops at various animal rights conferences and will continue to do so. We also plan to create video tutorials, host webinars, and provide free online workshops to spread this knowledge further throughout the movement, sharing strategies for improving AI accuracy, cost-effective access to tools, and data privacy solutions. We also recommend animal advocacy organizations visit NFPS.ai for a range of tutorials and other resources covering AI implementation for nonprofits.

   2. Personalized Consultations: Offering free consultations can support organizations in effectively integrating AI into their advocacy efforts. Open Paws has provided over 50 free consultations and will continue to provide these consultations to support organizations in effectively integrating AI into their advocacy efforts.

   3. Community Building: Additionally, creating groups or communities where organizations and advocates can learn together and from each other can increase AI adoption across the movement. Hive, an online hub for farmed animal advocates, is an excellent example of this, with multiple Slack channels dedicated to AI in animal advocacy.

2. AI Implementation as a Service: Provide customized AI solutions tailored to the unique needs of animal advocacy organizations.

   1. Open Paws has already implemented successful AI solutions for various organizations. For example, we developed an AI-powered automation for an organization that monitors the internet for misinformation, drafts responses, and syncs them to a database for expert review. Likewise, there are organizations like NFPS.ai that also offer AI implementation services for nonprofits.

   2. We will continue offering free AI implementation services to other animal advocacy organizations, providing tailored AI-powered solutions to enhance their operational efficiency and impact. This includes automation of routine tasks, content creation, predicting campaign outcomes, improve digital advertising performance and more, all designed to support and amplify animal advocacy efforts.

3. Developing Custom AI Models & Tools: These models and tools should include user-friendly interfaces for non-techincal users, prioritize accuracy (reducing hallucinations), be made freely available (to address budgetary concerns), be released with open source licences (so that organisations can host the tools locally to address data privacy concerns) and should also be free from speciesist bias.

   1. Open Paws is directly addressing these concerns through building a database of reliable information on animal advocacy, which can be used both to ground the responses from current AI systems in factual information (reducing hallucinations) and also to train custom AI models that we will release with open source licences (addressing budgetary and data privacy concerns).

   2. Studies show that retrieval augmented generation (RAG) using such databases dramatically decreases hallucinations from large language models, whilst the GraphRAG approach reduces hallucinations even further through the use of a database with both vector and graph functionality. This allows the database to better capture the complex relationships between different pieces of information.

   3. The database we are building is inspired by GraphRAG and uses this same vector-graph hybrid approach. This helps ensure that the AI models we develop are based on factual, up-to-date, and ethically-aligned information related to animal advocacy.

   4. Every AI tool and model we release (including this database) will be made freely available in web applications with user-friendly interfaces designed specifically for non-technical users. This ensures that our tools are useful to all animal advocates, regardless of their level of technical expertise. In addition to hosting these models and tools on our own easy-to-use platforms, we will also build integrations with popular software and applications, such as WhatsApp and Slack, as well as integrations with popular AI models, such as custom GPTs for ChatGPT.

4. Funding Assistance Programs: Given the severe lack of funding for AI initiatives in animal advocacy, we recommend funders create and promote funding assistance programs to address financial barriers to AI adoption.

   1. Grants for Technology Development & Adoption: Encourage funders to establish grants specifically aimed at helping animal advocacy organizations develop, adopt and integrate AI tools and technologies. Despite the potential of AI to significantly advance animal advocacy, this area has received minimal financial support, with organizations like Open Paws currently being run entirely by volunteers, without the budget to hire any staff members.

   2. Discounted Services from Technology Companies: Suggest partnering with technology companies to offer discounted AI services and tools to animal advocacy organizations, making them more accessible and affordable. This is crucial, as the current lack of funding means many organizations cannot afford these technologies without additional support. For this reason, our sister organization VEG3 offers a 25% discount to all nonprofits and shares hundreds of thousands of dollars worth of free cloud computing credits with Open Paws. We recommend that the animal advocacy movement also takes greater advantage of similar discounts from larger technology companies, such as the Azure nonprofit program which entitles nonprofits to $2,000 worth of free cloud computing credits per year, which can be used to access state of the art AI models like GPT4.

   3. Support for New and Emerging Organizations: Recommend that funders prioritize financial support for new and emerging organizations operating at the intersection of AI and animal advocacy. This support is essential to foster innovation and ensure the sustainability and growth of pioneering efforts in a field that has received limited funding. Given the nascent stage of the AI-animal advocacy intersection, we suggest that funders concentrate on providing smaller seed funding grants to a broad array of organizations. This approach allows for experimentation and learning from trial and error, rather than allocating larger grants to a more limited number of organizations.

By implementing these recommendations, animal advocacy organizations can harness the power of AI to enhance their effectiveness, optimize their efforts, and achieve their goals more efficiently. The future of animal advocacy can be significantly shaped by embracing and leveraging AI, ensuring that the future of this powerful technology benefits all sentient beings.