We commit to using AI in a way that helps us improve our internal processes while protecting company and client data, minimizing bias, respecting intellectual property, upholding ethical design principles, and minimizing environmental harm.
If a problem arises, AI is never an acceptable scapegoat. PixelSpoke is ultimately responsible for the final work product, even when AI is used.
We also understand that we’re still in an exploratory phase and when issues do surface, we will work collaboratively to find solutions.
All AI-generated output is reviewed, understood, and edited to meet our high-quality standards. We also verify that AI content is backed by credible sources.
If any problems are identified, we encourage team members to share their learnings so we can avoid similar missteps moving forward.
We strive to use AI in a way that enables us to focus more on the tasks that we enjoy.
Our goal is for team members to continue to feel a sense of ownership over their work, learn new things, and feel intellectually challenged.
As we continue to explore and experiment with AI, we encourage team members to be vigilant about identifying where AI may not actually be helping us.
When we default to AI, we sometimes spend more time refining prompts and editing the outputs than we would have spent simply doing the original task.
We are transparent with clients that we use AI to assist us with some tasks. We are also diligent about communicating our commitment to human oversight.
Using AI for any illegal, unethical, or malicious purpose.- Fully automating communications to clients, team members, and partners or other stakeholders without human review.
- Generating content that is discriminatory, harassing, defamatory, inaccurate, or misleading.
- Automating decision-making in areas that impact fairness, access, or equity without any human oversight.
- Inputting any client or company confidential information into public AI models such as NPPI data (non-public personally identifiable information), medical records, account information, or any other private member data.