Introduction:
Devshi Mehrotra and Leslie Jones-Dove were motivated to create technology focused on the needs of populations historically disadvantaged by law enforcement while studying computer science at the University of Chicago. They AI to Transcribe Evidence for Public Defenders, who informed them that the hours’ worth of jail calls, body cameras, and other recorded evidence overwhelmed them.
One estimate states that a body camera used by an average police officer will record around 32 files, 7 hours, and 20GB of video per month in 720p resolution. These numbers easily mount up when multiplied by the hundreds to tens of thousands of police officers in the police force.
In an email interview, Mehrotra said, “On the one hand, body cameras and other gadgets are vital for holding law enforcement accountable and giving the best defence possible. However, they make it harder for public defenders to handle their caseloads, which are 3 to 10 times acceptable.
As a result, JusticeText was formed by Mehrotra and Jones-Dove and is one of the 200 firms competing in the Startup Battlefield at TechCrunch Disrupt 2023. JusticeText enables public defenders to capture time-stamped notes, produce video clips, AI to Transcribe Evidence for Public Defenders and exchange evidence with their colleagues while automatically transcribing body cam footage, interrogation videos, and more.
Public defenders must provide the best defence for their clients. However, the assessment of digital evidence consumes an increasing amount of scarce resources, according to Mehrotra. Saving time on the discovery review process frees up valuable resources that can be used to work on the case and cultivate client relationships. Fewer IT resources will be used to solve technical issues, such as figuring out how to play movies in special proprietary formats or producing video evidence for a trial from a technical standpoint.
AI to Transcribe Evidence for Public Defenders:
AI to Transcribe Evidence for Public Defenders (Image Source: techcrunch.com)
While that sounds great in theory, some public defenders might be hesitant to upload evidence to a website like JusticeText. One issue is that the evidence may be made vulnerable to data breaches. Additionally, lawyers may have legitimate concerns that uploaded data may be utilized in ways neither they nor their clients necessarily consent to, such as developing JusticeText’s AI models.
I’m also concerned about JusticeText’s quality of transcriptions, especially because transcription technology sometimes AI to Transcribe Evidence for Public Defenders accurately translates accents and languages. Considering that nuance isn’t precisely AI’s strong suit, JusticeText’s summaries may also need to be more accurate.
Mehrotra made an effort to ease my concerns by stating that JusticeText only produces a summary for transcripts that are more confident than a specific level and that it offers users a chance to redact personal information from transcripts. JusticeText stores data in “secure cloud servers” and encrypts it in transit and at rest. According to Mehrotra, JusticeText doesn’t use uploaded data for model training unless it gets specific permission.
Since JusticeText’s $2.5 million funding round was completed (with participation from Bloomberg Beta, True Ventures, LinkedIn co-founder Reid Hoffman, and former Stockton, California, mayor Michael Tubbs), it has increased the number of public defender offices that use its services, now including the statewide public defenders systems in Massachusetts and Kentucky.
According to Mehrotra, JusticeText now partners with over 100 public defender offices, charitable organizations, and private criminal defence firms nationwide, generating $1 million in recurring revenue annually. Since we last met with JusticeText (in September 2022), when the business had between 50 and 60 relationships, deal activity has considerably increased.
Shortly, according to Mehrotra, JusticeText will concentrate on adding a Spanish-to-English translation feature and support for “multi-language” recordings, such as those that contain speech in both English and Spanish. (I hope the team takes caution while implementing this because AI isn’t a flawless translator, either.) In the coming year, JusticeText also intends to expand its staff from its current seven members to approximately ten, focusing on strengthening its marketing and communications department.
According to Mehrotra, the pandemic has harmed our end users by causing a backlog in the criminal justice system. Even while critical decision-makers could be harder to reach in light of these difficulties collectively, this creates a flaming platform for change.
