Thread by @techreview

Welcome to #EmTechDigital! Our signature AI conference is back, in-person and online. We'll have 30+ speakers over the next two days. Join us and grab a virtual ticket here: trib.al/YipvRIp

In our opening keynote with @AndrewYNg of @landingAI, we’re going to address a decade-old legacy, that algorithms can do remarkable things. A focus on better data promises to bring the benefits of AI to a wider range of people, and it may help make AI safer and more reliable.

@AndrewYNg @landingAI "We should collect more data." - @AndrewYNg #EmTechDigital

@strwbilly: How do you standardize data curation?

@AndrewYNg says it's early in the development of data-centered AI, but his team has been working on tools to recognize discrepancies in labeling. #EmTechDigital

Education and tools will be key to data-centric AI. #EmTechDigital

"There's a lot of room for innovation in storage." - @AndrewYNg #EmTechDigital

@AndrewYNg We've been feeding AI a lot of junk food, but we can engineer high-quality food for AI too, @AndrewYNg says. #EmTechDigital

@AndrewYNg What skills should be developed to be successful for data-centric AI?

Data science skills will be important for a long time, Ng says. But keep adding to those skills. #EmTechDigital

Moving to our next speaker, we're hearing from @rajcs4 of @snorkelai on the mechanics of data-first AI. #EmTechDigital

Data labeling is key to determining the success or failure of AI applications. @rajcs4 is covering how implementing a data-first approach can transform AI inference, resulting in better models that make better decisions. #EmTechDigital

@rajcs4 Data-labeling takes time. But programmatic labeling turns labeling into code. #EmTechDigital

@rajcs4 Collaborating with subject matter experts can improve programmatic labeling and a model's overall performance, Shah says. #EmTechDigital

@rajcs4 Programmatic labeling can dramatically accelerate the speed of AI development, @rajcs4 says. #EmTechDigital

@rajcs4 What does the approach change? What does it make possible that wasn't possible before?

@rajcs4 says it creates collaboration between subject matter experts and data scientists. #EmTechDigital

@rajcs4 Is synthetic data an extension of programmatic labeling?

Absolutely. It's compatible, Shah says. #EmTechDigital

@rajcs4 We’ve just heard from two great speakers about how better data leads to better AI. But “better” isn’t just about the quality of your labelling, or data curation. It’s about ethics too. AI is not going to be widely and reliably used until we get this right too. #EmTechDigital

@HaniyehMah is the Global AI Ethicist at @DataRobot. Thinking about ethics should happen at every stage of a project, she says. #EmTechDigital

If you haven't considered ethics, you might end up with damage to your reputation. Users need to trust the process, @HaniyehMah says. #EmTechDigital