How IP.com Is Ensuring Ethical AI

IP.com’s suite of solutions for innovative businesses is backed by our proprietary AI engine. AI allows us to develop tools that help companies understand where and how their competitors are innovating, find white space within the patent landscape, and bring novel ideas to market faster. Yet, with this power comes responsibility. AI’s machine learning algorithms have a reputation as a mysterious black box. As explained by researcher Samuele Lo Piano in a 2020 journal article, the quest to maximize “usability and effectiveness” may be perceived as the sacrifice of interpretability, as well as “fairness, accuracy, accountability, and transparency.”

However, IP.com prides itself on implementing ethical, transparent AI while still empowering organizations with intuitive access to global patent and litigation data.

What is ethical AI?

As a society, we rely on AI to uncover insights and make important decisions about hiring, lending, healthcare, driving, and more—including business strategy around innovation. When AI becomes part of the decision-making process in industries like human resources, finance, policing, and patenting, ethics are of the utmost importance. The Harvard Gazette, in a 2020 article titled “Great promise but potential for peril,” offers three broad areas of concern:

• Privacy and surveillance,
• Bias and discrimination, and
• The role of human judgment.

Responsible AI respects individual rights to privacy and nondiscrimination without manipulation. AI can’t determine if it’s ethical; it’s up to humans to ensure the AI they’ve created and trained is (and remains) ethical.

It’s important to note that unethical does not mean illegal. AI can operate well within current laws while failing to meet ethical standards for the industry. For example, Facebook famously operates in this grey area. The information you share with Facebook, both directly and indirectly, shapes your experience within the platform.

How does IP.com ensure ethical AI?

In contrast, our AI engine does not manipulate source data to advantage or disadvantage our users in any way. While a user could encourage a specific outcome with their actions within the software suite, the platform cannot be manipulated to influence outcomes not prescribed or known by the user beforehand.

Shane Saunderson of the University of Toronto poses four questions to help users of AI-backed systems identify transparent, responsible algorithms:

1. Who owns it or is the interested party behind it?
2. What are its objectives?
3. What tactics is it using to reach those objectives?
4. What data does it have available?

When you use our tools to drive innovation forward, you can answer each of these questions with confidence. IP.com is a US-based, ITAR-compliant company committed to helping innovative companies evaluate, protect, and monetize ideas. Our AI engine—trained on high-quality technical literature and litigation data—is in inference mode, constantly improving its ability and accuracy.