In two and a half years, this series has never focused on AI! That ends today.
We have updated our AI principles and developed an AI vision statement. I'll dig into those today, as well as encourage you to create an AI policy if you haven't yet—using TAG's awesome Responsible AI Adoption Framework as your guide.
Submittable's AI Vision Statement
We recently went through the exercise of creating an AI vision statement that reflects the future that we're building toward. We hope it's one you are excited to be a part of:
Our vision is to make AI our customers' trusted partner in progress.
Submittable’s customers do fundamentally human work tackling society's greatest challenges. They steward public and private resources to feed the hungry, serve vulnerable populations, and safeguard our planet for the health and enjoyment of future generations.
AI tools built for this work can—and must—help us rethink what’s possible while preserving what makes us uniquely human.
Through strict human oversight, we will create AI tools that go beyond simply helping our customers work faster. Our customers will gain essential partners in their work, solving complex problems more completely and improving the lives of people around the world.
By getting it right, we help AI answer the call to advance the best of humanity.
Submittable's Responsible AI Principles
Our Responsible AI Principles guide our development of AI in our product suite. Submittable first released our Responsible AI Principles in late 2023. AI changes fast, and the average person's understanding of and engagement with AI has changed a lot since then (my hand is certainly up!).
Plus, with new leadership of our product and engineering teams following Submittable's acquisition of WizeHive last year, it was time for a fresh look at the principles guiding our use of AI in product development.
This refreshed set of principles is not at all a departure from our original set, but it does reflect a stronger send of caution, as can be seen in the concepts of reversibility and responsiveness.
Focused & Empowering: Help people achieve more within tasks explicitly defined by humans.
Transparent & Controlled: Make users aware of when, where, and why AI is available, and allow them to choose whether or not to use it.
Accountable & Reversible: Maintain human control at all times. Every AI-influenced action is traceable, reversible, and responsive to human intervention.
Equitable & In Community: Prevent AI from perpetuating or amplifying systemic inequities by mitigating and correcting for bias, limiting its decision-making scope, and consulting community stakeholders.
Private & Secure: Implement world-class measures to ensure data is private, secure, and never shared.
Why You Need an AI Policy
Does your organizaiton have an AI policy? If not, here's why you need one.
It only takes one employee engaging in AI foolishly or recklessly to compromise sensitive data. Picture this: someone enters confidential grantee data (such as financial information or program performance) into a free AI chatbot for something theoretically innocent, like help rewriting an email. It's possible that information could be used to train the model, meaning confidential details could surface in responses to other users, or other exposure like being accessed by the provider.
Beyond the immediate risk or breached confidentiality, you've now potentially violated agreements, created legal liability, and risked your reputation.
Sound scary? It is! But a clear policy that everyone understands (and understands why it matters) is your first and best line of defense. To get started, I recommend you check out the Technology Association of Grantmaker's simple (and free!) Responsible AI Adoption Framework.
TAG's Responsible AI Adoption Framework
TAG's free Responsible AI Adoption Framework is designed to help funder's center their values as they bring AI into their work. As Jean Westrick, Executive Director of TAG, wrote in a recent call to action for the sector: "Every organization needs an AI policy that centers their values and creates clarity for staff about how AI can be used and how it shouldn’t."
The framework is divided into three core areas of consideration:
Organizational considerations: Awareness & engagement; Training and change support
Ethical considerations: Resonsibility & Acceptable usage; Privacy, bias & transparency
Technical considerations: Implementation, oversidht, adaptiation; Vendor evaluation
And, it explores use of AI across four avenues:
Individual use, such as using Chat GPT to help draft emails or take meeting notes
Organizational use, such as AI use in business tools like Submittable
Mission-related use, such as measurement and evaluation across programs
Nonprofit Enablement use, such as supporting fundraising or prospecting for grantees
I'll call it there for today. I hope you find these helpful! We'll be taking a break from Best Practices until the new year, but I'll see you in our regular Product Roundups before then.