Which complicated technical challenges need solutions immediately? RealLIST Engineers on cybersecurity and biased AI

As an industry, tech is growing at a faster pace than ever before, presenting the world with disruptive innovations and new problems to solve.

In a recent conversation with Technical.ly’s 2021 RealLIST Engineers on our community Slack, we asked top technologists from all five of our markets what thorny technical problems they’d like to see solved ASAP — and why that problem’s been left unsolved so far.

Join Technical.ly's Slack

Alison Alvarez, cofounder and CEO of Pittsburgh’s BlastPoint, said that she’d like to see more technologists and companies treating cybersecurity “like epidemiology” by acknowledging the real and ongoing risks the threat poses to society. (Hello, Cybersecurity Awareness Month!)

“The general public does not have the awareness or tools to protect themselves, and neither do many corporations,” she wrote. “The basic principles should be treated like hand-washing, as in we should be talking about them early and emphasizing them as a part of life. The methods should be tweaked as technology changes — hey, now there’s hand sanitizer. We also need cheaper and more robust auditing principals so there is no excuse not to adopt.”

Others in the conversation echoed Alvarez’s concerns, including Matthew Ford, the principal architect at Baltimore-based Protenus: “I used to work in critical infrastructure and the thought that ‘we’ll slap some security on at the end’ was and is terrifying,” he said, agreeing with Alvarez that the response should be more thorough throughout the software development process.

Jean Lange, a senior technical program manager for LendingHome, offered exceptions to Ford’s and Alvarez’s points, of where security concerns might not always be as relevant.

“The thing about this is that there are some places where security doesn’t matter, and some places where it matters all the way, and you’ve got to treat them as themselves and not each other,” she wrote. Lange added that examples of this might be personal projects for learning, informational websites for businesses, and organizations or projects that have personal use but are kept private from others. Still, she said that it’s a spectrum that requires risk analysis in each case.

Cybersecurity concerns have become increasingly widespread with the onset of recent attacks and the number of people now using technology for much of daily life during the pandemic. Even social media outages, like the one seen by Facebook earlier this month, signal a need for better security that the RealLIST Engineers are seeing in their day-to-day work.

Brandon Coates, a DC-based senior cloud security engineer at Yahoo, shared a glimpse into what he comes across in his role.

“As a security person I endorse this message,” he wrote in response to Alvarez’s initial answer. “However we also have to look at our tooling. I tried to get family members to use a password manager and sometimes it’s more confusing to use that then just writing down the passwords (depending on threat model might be the safest IMO).” In other words, creating security that’s easy to use and integrate will be essential to its success.

Another common answer to the prompt involved the need to mitigate bias as artificial intelligence and technology based around it continues to advance.

“As more and more things interact with an AI/ML system, this is going to continue to be a HUGE HUGE impediment,” Ford wrote. He added that while he himself doesn’t know what the solution looks like, others he’s spoken with in the tech community have similar concerns.

Lange shared her approach to this problem in software development at an individual level.

“My tiny personal contribution to this problem is to try to be the kind of human I want the robots to learn from in spaces where robots will learn from me,” she said. In another thread, she and Ford traded notes about cyborg systems — technology that merges both robot and human aspects — as a potential middle ground solution.

Baltimore-based MindStand Technologies cofounder and CTO Eric Solender offered other pathways to addressing bias.

“As everyone knows, AI is all about data, and if you have biased data you will have biased models,” he wrote. To prevent that, the teams behind writing the code, training the models and more should have people from a wide range of backgrounds who can offer diverse perspectives. “If you have more diverse input into what these models learn from, theoretically the output will also be more diverse. This is by no means an all encompassing solution, but I think it goes in the right direction.”

To read the entire conversation, check out the #ama channel in Technical.ly’s community Slack:

Join Technical.ly's Slack Sophie Burkholder is a 2021-2022 corps member for Report for America, an initiative of The Groundtruth Project that pairs young journalists with local newsrooms. This position is supported by the Heinz Endowments.