SOSA, an Israel-based open innovation company, has released a report on how to tackle some of the upcoming challenges posed by Artificial Intelligence. The report, titled ‘XAI: Are We Looking Before We Leap?’ highlights the challenges and advances when it comes to regulations, use-cases, and emerging technologies ‘taking the mystery out of AI.’
“I think it’s important for everyone, the public in general, to understand what they don’t understand, or what no one really has pushed transparency on,” Uzi Scheffer CEO of SOSA explained. “It’s important for entrepreneurs and for the tech ecosystem to understand that maybe if you’re targeting some sectors, it’s probably a good idea to integrate these guidelines that are not standardized.”
A yearning for better communication surrounding AI and how it is deployed by companies has recently emerged in public forums in a big way. The study credits Netflix’s ‘The Social Dilemma’ for bringing the discussion to the mainstream and the world realizing some of the potentials that this technology has over everyday lives. One of the ways to tackle this ‘black box’ around the information is to improve the transparency behind the technology and its capabilities.
“With transparency, you can provide accountability, allow auditability, and there is a meaning to liability because things are traceable,” said Scheffer, adding that SOSA has seen a “great rise in the adoption of Explainable AI.”
Explainable AI, the techniques of adopting AI technology that can be best understood by humans, is addressed in four major sections of the report. The first, which focuses on the regulatory actions that can be taken by governments, highlights Europe, the US, and China as the main continents that can help steer governing policies like privacy, data control, and diversity and non-discrimination actions.
“Nothing is standardized on a global level yet, but if you ask me what my estimation is, it’s that it will follow the data privacy regulations journey,” predicted Scheffer. “GDPR, California, New York, if you read them and the details, then they’re all quite similar but not standardized yet. Whether they will be standardized on a global scale, is an important question.”
Some private companies and corporations are already making what the study calls ‘notable strides’ to help humans understand the decisions made by AI. The second part of the study highlights companies like Deloitte, Intel, and Wells Fargo, where transparency, accountability, and trustworthiness of data-driven AI decisions are essential. Deloitte’s tool, Lucid ML, for example, provides customers with a better understanding of the Machine Learning models and their inner workings when identifying its risk resolutions.
“Lucid, Deloitte’s tool is great and it’s a nice example… mankind made sophisticated technologies and implemented them, and now there is a race to stress test them, understand them better, and model them,” Scheffer explained.
Part of SOSA’s mission is to help benchmark, identify opportunities, set innovation strategies, and find the solutions to some of the biggest problems in the tech ecosystem. Sitting ‘at the intersection between supply and demand’, SOSA can help highlight companies that are specifically trying to take the mystery out of AI. Out of the eight companies the company highlighted in the report, two are Israeli - Superwise.ai, which helps decrease data science teams’ workflows and labor costs; and Mona Labs, which is an AI-based algorithm quality check system.
“The companies in the study are here because they offer a good overview on what is being developed in the tech ecosystem,” Scheffer explained. “We highlighted companies that are already offering solutions and touching on these emerging subjects of Explainable AI - but we don’t have a formal business relationship and we’re 100% unbiased.”
Introducing legislation or public awareness around some of the newest technologies in our lives is no easy feat. Often, the law moves at a snail’s pace compared to the ‘move fast and break things’ attitude of tech companies. However, SOSA believes that today, there is a lot more awareness of the problem and challenge of understanding computer-based decision-making. And while strides are being made in protecting our privacy, it makes sense that the mystery of AI will be solved just as fast.
“Unlike other challenges, I think this will be sorted out quite fast, maybe faster than the data and privacy,” Scheffer concluded.
Originally Published on CTech