Legal Robot will publish this report quarterly, the next being on or around July 1st, 2018.
We will make our owners, designers, builders, users, and other stakeholders of analytic systems aware of the possible biases involved in their design, implementation, and use and the potential harm that biases can cause to individuals and society.
In an effort to improve the general awareness around Algorithmic Transparency, our CEO, Dan Rubins, traveled to Washington D.C. to speak at the Association for Computing Machinery’s (ACM) Panel on Algorithmic Transparency. The panel discussed the challenges, opportunities, business value, and societal impacts of algorithms with a diverse and lively crowd of political staffers, lobbyists, academics, and other stakeholders.
We will adopt mechanisms that enable questioning and redress for individuals and groups that are adversely affected by algorithmically informed decisions.
Most predictions in our app have a button to visualize and examine the details of the result, however we don’t provide this for basic operations like sentence segmentation, part-of-speech tagging, and other NLP operations that are fairly well understood by the NLP community. Where appropriate, we also include statistical measures like precision, recall, and F1 score, as well as the size, source, and scope of the underlying dataset, and details about the design of the algorithm used for the prediction. Of course, we don’t expect everyone to be able to interpret this technical data, so we also allow anyone to share the results with our team for more explanation.
Also, anyone can ask questions over email to [email protected], even if they are not using Legal Robot. These questions are tracked separately from our normal support requests.
We will demonstrate to our users how decisions are made by the algorithms that they use, even if it is not feasible to explain in detail how the algorithms produce their results.
Many of our processes at Legal Robot use deep neural networks to process language. Neural networks can be very complex which can make them seem incomprehensible. However, just because an algorithm seems like a black box (and is treated that way by many people using it) does not mean it cannot be explained.
To begin with, we do not use any 3rd party machine learning APIs at Legal Robot. This is mainly so we can control where data processing occurs. Rather than passing sensitive data to a 3rd party as many “AI” companies do, we actually build our own algorithms so we can open up the internals for further analysis and explanation.
We now tag each prediction created by our software with a unique random identifier that can be used to trace back to both the algorithm and the training dataset used for each prediction in order to enable questioning and redress.
We will produce explanations regarding both the procedures followed by the algorithm and the specific decisions that are made.
Some of the techniques we use yield dense vectors (basically a long string of seemingly incomprehensible numbers, like [0.78524 , 0.42504, 0.60494, …]) that we use to teach an algorithm what a particular type of clause looks like (statistically speaking). However, we are working on methods to make these dense vectors more interpretable, much the same way that deep learning techniques can yield semi-interpretable layer visualizations in computer vision. We think these can provide some utility for users to understand what is happening inside the “black box.” We are focusing on these areas over the next few releases and intend to publish our results to the research community.
We will provide a description of the way in which the training data was collected, along with an exploration of the potential biases induced by the human or algorithmic data-gathering process.
Every model created by Legal Robot is traceable to the specific dataset. Every data point also includes detail on how and why each sample was collected, and the details of any enrichment or manual tagging.
All of our models, algorithms, and datasets are now versioned and recorded, providing a full audit trail. We have not yet set a policy or provided a mechanism to view or download the audit trail, but are planning to release this feature soon.
We will use rigorous methods to validate our models and document those methods and results. In particular, we will explore ways to conduct routine tests to assess and determine whether the model generates discriminatory harm. We will publish a description of the methods and the results of such tests in each quarter's transparency report.
We are working on a structured approach to analyzing bias to capture both known and unknown biases. In addition to this high-level approach, we are investigating lower level techniques like attribution to detect and evaluate bias. Last quarter, we started to use automated bias analysis on some of our models, but there is still much work to do by the research community.
We had submission of new reports to our bug bounty program on pause for much of this quarter as we redesigned many of the core functions of Legal Robot and pushed new features related to GDPR. However, our team continued to work on existing bug reports; a real time feed of the activity is available at hackerone.com/legalrobot
We intend to disclose all reports, once closed. However, we also respect the wishes of security researchers that are working with other organizations to resolve related issues.
We require all members of the Legal Robot community to abide by our Code of Conduct. As of the date of this report, we have not received any reports alleging violations of our code of conduct.
For more information around what inspired this statement go to https://www.canarywatch.org.
As of April 2nd, 2018:
Special note should be taken if this transparency report is not updated by the expected date at the top of the page, or if this section is modified or removed from the page.
The canary scheme is not infallible. Although signing the declaration makes it difficult for a third party to produce this declaration, it does not prevent them from using force or other means, like blackmail or compromising the signers’ laptops, to coerce us to produce false declarations.
Legal Robot has not received any “take down” notices or other removal requests under the Digital Millennium Copyright Act (“DMCA”) or any other regulation like Article 12 of Directive 95/46/EC, or the newer Article 17 of the General Data Protection Regulation (“GDPR”), commonly known as the “right to be forgotten”.
The news quotes below show this report could not have been created prior to April 2nd, 2018.
-----BEGIN PGP SIGNATURE----- wsFcBAEBCAAQBQJaSzV4CRCY0PbwMF7zeAAAlq4QAAFofc+0rQK7FpEeVmqSAst+ LcKJJ5kkd0Aktsl9PQtvdUdyAy+3iLY41a7lKWuTb09xRQX9UreIFzdDUOoZWomx wFGwCv4ru5/Yb4VfFMcwSGYtlydBTnYwsF0y9WF6gyn7eEXA8RtIh7Rb52vOGE0p qdmJOzx7jS3hCKKSMxibhs8KQNI3ohU7NfpZiRKdW8Tn64xbGoUXpC/cGdcBvoI9 t4pNq3rIkPB4gJOQrgC5O4p6u9WPxODAe8S3XZ4H033WTiw08Y2grmqULnnF5eeq F/vk6dXJMAeLfEV6SiIYY3GNx2v6RzjLJeJjOdFxMAyUAgEL4/TLbfdYJslnTzv2 8TnB62WInRsI2p6iGd2F5dzkhTFXHvgUM8Oo6iqmWCMbYqkA9ZUoaC6gVJO2UndL jM4eD0VHWMUBBtU6UFKz6vZ9/a5p0MTIqKvK3FUhnUFwQslcKmWqtHOYPImsjdN3 tmffO7Iu2WXhCIIljTEemo1/CdnEPcrl0/xGNpHdXFFF3KZpCknDLR7iGK0SSnPL 0a2ni+v5UUupulrMPJSU7eVOKS+ipcOswZG+HLmjKmhPwHnGnanP8aTKjiWYH1KT lU3rGVIVW/gaGg2b30EhBKOL/eV1/nQ0tw0ZAnwzfaEsFIrvmxyue4q68is20PFq 8HMkOodgKdxYBaFaCvWF =Z+Zu -----END PGP SIGNATURE-----