Up Next

Only half of students with disabilities ask for accommodations, Verbit says

Back

Video is the most common way higher education institutions engage students with distance-learning content, but few caption all their video content for accessibility, according a survey being released Tuesday by the transcription and captioning company Verbit.

About 14% of schools reviewed by Verbit provided captions as a default, while about 10% said they only caption lessons when a student requests it. About 76% of content is partially captioned.

But with just 52% of students interviewed saying they would ask for accommodations like captioning if they’re not automatically provided, Verbit concluded that colleges and universities may need to be more proactive in making learning materials more accessible.

“This lack of reporting and transparency is critical to consider when aiming to craft inclusive classrooms or environments, as its likely to emerge as an issue should school leaders not proactively address or offer such accommodations to the greater student population,” the report reads.

The U.S. Department of Education recommends institutions review, with their faculties, how to use built-in accessibility features like captioning or transcription on videoconferencing platforms like Microsoft Teams, Zoom and Google Meet. These platforms offer built-in captioning, but federal guidance recommends reviewing any transcriptions to make sure they meet accessibility guidelines. Of the schools Verbit surveyed, about 85% plan to offer courses in online or hybrid formats in the upcoming academic year.

About half of schools record lectures and transcribe them later, while about 44% provide live captioning and transcriptions, the survey found. The report also showed that accommodations were less likely to be used in settings outside the classroom but that are still part of the college experience, like live talks.

Nearly 30% of school employees told Verbit that live captioning and other accommodations — including sign-language interpreters, audio descriptions and foreign-language translations — were offered at university events, with 12% making them available at sporting events.

Verbit recently said it plans to make an initial public offering in 2022, after acquiring rival captioning provider Vitac, the country’s largest closed-captioning provider in media and entertainment. Other companies are also looking to provide accessibility tools to higher education such as Blackboard’s “Ally,” which audits online content for accessibility issues. In a recent assessment by the company, a majority of the files checked by the software contained “severe to major accessibility issues.”

The full article can be read here.

Up Next

AI-Powered Legal Tech Isn’t All Automation—or Even Machine

Many legal tech platforms tout their advanced artificial intelligence capabilities. But behind that software is often a long-standing reliance on human reviewers verifying results before they are seen by the end user. While the dominance of such a hybrid review model may be waning, observers said that the combination of AI and human review will likely always have a place in legal tech.

For some in the legal profession, such an approach strikes a necessary balance of automation and trustworthiness.

“In the legal space, it’s more important,” said LexisNexis chief data officer Rick McFarland. “The professional-grade space where we are dealing with legal or medical [data], I think it’s more important to have [human and AI review] than if you’re in the consumer space.”

Indeed, even when transcribing legal proceedings, more professionals are embracing automation. But they still expect a human to verify an AI-produced transcription, noted Verbit legal strategist and customer success manager Anthony Sirna. Verbit provides legal transcriptions in real time with AI and human reviewers to verify the transcription’s accuracy.

“You still are going to have the human element,” Sirna said. “I think in particular with transcriptions, having another set of ears and people that understand the legal language, you want people to catch [mistakes] and make it as accurate as possible.”

Human review also has a place in legal research, as some research platforms delve into the granularities and unique insights of a court action. Lex Machina, for instance, leverages both software and attorneys for assessing damages, findings and other nuances, noted Lex Machina product management director Carla Rydholm.

“For some data sets, yes, absolutely we use a machine-assisted human review approach for high-value data derived from analytical thinking about what happened to legal claims brought in a case,” she said. “Attorney review is important for judge orders to analyze the claims brought and what happened to those claims on the merits, and to verify the practice area that the case belongs to.”

The prominence of such a hybrid approach won’t likely end any time soon, noted HBR Consulting legal technology practice senior director Rohit Gulati. “I think in the future I can see where pieces of the work are fully automated, but given the type of work legal professionals do, which is a lot of interpretation of [matters], I don’t think the AI tools are quite there yet where they can take on the interpretations,” he said. “I do think there’s going to be a hybrid model for a bit.”

But while the reliance on combining human and AI review is prevalent in legal tech, some legal tech providers aren’t explicit about that process in their software, Gulati noted.

“The vendors want to focus on their AI technology and say how robust it is, [while] some vendors are trying to give their clients the peace of mind of: There is technology looking at [an invoice’s] line items but there’s a secondary review going on. It’s a mixed message from vendors and it goes to the level of sophistication they want to market to clients,” Gulati said.

However, as AI becomes more accurate, legal tech providers will look for ways to leverage less manual review while maintaining quality, noted LexisNexis’ McFarland.

“I do think it’s a cost model that has to be considered. There’s a sweet spot you have to find. You don’t need a human to check everything or a human checking one thing. That’s one of the challenges we’re [finding] of where’s the optimal sweet spot where we feel we’re doing our due diligence at the quality we’ve set without hiring 10,000 lawyers for [quality control],” McFarland said.

Read the original article here.

Back To Top