June 24, 2019
Just last month, we had a great opportunity to be part of a larger conversation around AI standards.In particular, we responded to this request for information (RFI) from the United States National Institute of Standards and Technology (NIST) to create a plan for Federal engagement that supports reliable, robust and trustworthy AI technologies via a common technical standard.
At Neo4j, we strongly believe – and I personally believe – that artificial intelligence should be guided not only by robust technical standards, but robust ethical standards as well. A key overlapping principle in both of those areas is context.
Contextual information not only results in better performing AI systems, but also in a clearer ethical perspective for those creating and shaping it. For example, context helps us understand the factors and pathways of logic processing (and be able to explain them) so we can hold organizations accountable for AI decisions.
Context – in both data and in life – is derived from connections, and what’s better at natively storing, traversing and analyzing connections than graph technology? Nothing.
So in response to the NIST request, last month we submitted an open letter – co-written by myself and my brilliantly talented colleague Amy Hodler – on why graph data technology should be considered as part of a technical standard for AI solutions.
Here’s a copy of the introduction we sent – details on the rest of the RFI are below:
–––
To whom it may concern,
Re: RFI: Developing a Federal AI Standards Engagement Plan
The case for graph technology as a foundation for AI
We are writing in response to the request for information from the National Institute of Standards and Technology (NIST) regarding creating a plan for Federal engagement in the development of technical standards and related tools in support of reliable, robust, and trustworthy systems that use Artificial Intelligence (AI) technologies: document 2019-08818.
The potential power of artificial intelligence is expansive and will be used in ways we cannot yet imagine. Despite, and perhaps because of this, we have a duty to guide its development and application in ways that facilitate innovation and fair competition, public trust and confidence, while incorporating the appropriate protections.
We are Emil Eifrem and Amy Hodler, the Chief Executive Officer and Analytics and AI Program Manager for Neo4j, Inc., a California-based company. We have been involved for years with data technologies that specifically deal with how people, processes, locations, and systems are connected and interrelated.
Neo4j helps people make sense of data through graph technologies which naturally store, compute, and analyze connections and pathways among data points. AI and machine learning systems are more effective, trustworthy, and robust when underpinned by contextual information provided by graph platforms. We can assist the Federal government in understanding the significant role that connected data plays in AI and learning systems.
Today, Neo4j leads the graph platform category in installations which include numerous commercial and Federal projects including the Department of Defense, the United States Intelligence Community, as well as state and local government agencies. The U.S. Army has deployed Neo4j for tracking equipment maintenance in their procurement process. MITRE Corp uses Neo4j for managing cybersecurity, and NASA consolidates and references its past research with a Neo4j powered knowledge graph.
The private sector has often been a leader in technical standards, but we also believe that public-private initiatives fuel innovation and assure transparency. We are major contributors to open source projects and support non-profit organizations including our work with NASA and the International Consortium of Investigative Journalists (ICIJ) on the Panama Papers, which 3 years on has resulted in more than $1.2 billion in tax fraud investigations.
Working with both highly as well as lightly regulated industries and governments, we’ve learned that complex data – and its use in AI – are a worldwide concern, which no single organization should regulate alone. We are offering our suggestions and support in your efforts to develop standards for AI technologies.
In summary, context must be incorporated into AI to ensure that we apply these technologies in ways that do not violate our societal and economic principles. AI standards that don’t explicitly include contextual information will result in subpar outcomes as solution providers leave out valuable, adjacent information. We have attached our recommendations for standards and tools to guide AI technologies such that they become truly reliable, robust, and trustworthy.
With our sincerest regards,
Emil Eifrem, Neo4j CEO & Co-Founder
Amy E. Hodler, Neo4j Analytics & AI Program Manager
–––
In the following weeks, Amy will go into further detail (her series starts here) – drawing upon the full report we submitted to the NIST – on how and why graph technology is the superior choice to provide ethical and economic context to AI solutions.
Whether through an official standards body or through marketplace-wide agreement, I hope that one day all artificial intelligence systems benefit from the ethical, economic and technical benefits of connected data context.
–Emil