(bright music) - [Narrator] Johannes Gutenberg invented the printing press around 1440, but his innovation wasn't
creating new technology. His new machine combined
existing print block technology from East Asia and mechanical
devices first developed by the Romans for making
wine and olive oil. Moreover, his innovation launched just as literacy rates were on the rise. Those events ignited an unparalleled age of
scientific discovery, empowering people to share ideas and information farther and faster. Textbooks, newspapers, and even the internet have their roots in Gutenberg's printing press, but even he couldn't have imagined the legacy his inventions would leave, for better or worse. Today, businesses are primed
for advancements in AI to improve how we work,
live, learn, and connect, so what can society learn from the genesis of Gutenberg's press as AI makes its mark on the future? (bright music) (screen whooshes) (bright music fades) (bright upbeat music) Like the Gutenberg press, AI is a tool.

and the end result of working with it depends on how it's used. To help explain, here's VP of IBM Research AI, Sriram Raghavan. Just as opening up the technology of printing fundamentally transformed knowledge and dissemination, and then society as we know it, I think we need to think of AI in the same way. $16 trillion worth of additional GDP going to be unlocked in the next five to six years. That's enormous in terms of how it's going to change the fabric of society. And the creation of communities to develop the next generation of purpose-built, trusted AI models is an important next step. By offloading repetitive tasks, AI offers businesses a chance to empower their teams, refocus efforts, and create value with their ingenuity. That is the potential of an open approach to AI. With generative AI models, such as Granite, and open-source projects like InstructLab.

enterprises can build AI on a foundation of trusted data by deploying custom models that speak to their specific business needs. IBM and Red Hat have created an open source project called InstructLab, and the core of InstructLab is an approach that allows AI models to be developed collaboratively, allowing the innovation ideas of the community to evolve more. We released recently a family of models from our Granite family of models out in the open with an Apache license that gives developers, academic institutions, startups, governments across the globe, highly performant, state-of-the-art models that they can then use and work with. And we are very excited, again, by what a combination of open base models, such as our Granite models, and an open approach to model development, how the duo can really supercharge open innovation in AI. [Narrator] Just as Gutenberg's innovation helped unlock great prospects for the world,

the printing press also had the potential to be used nefariously. Disinformation, propaganda, and libel are the flip side of the coin from poetry, textbooks, and journalism. Other technologies have suffered the same fate. Television, the internet, and smartphones have all required some form of regulations and safeguards to maximize their potential while minimizing danger, and AI is no exception. While IBM publishes the data sources that train its AI, many models are still trained on datasets of unknown quality or origin. AI industry leaders like IBM understand how crucial it is to use trusted, ethically sourced data, free from toxic input so enterprises can avoid bias, incorrect claims, and toxic output. This is why IBM, in collaboration with over 100 leading organizations across industry, startup, academia, research, and government, launched the AI Alliance to maximize safety, security, transparency, and trust in AI. Today, in the world of AI, there is a lot of focus on quality.

but there isn't a consistent approach to evaluating safety, fairness, robustness of the models. And a big focus area of the Alliance is to create the best set of benchmarks by which to measure and evaluate safety in AI models. - [Narrator] With collaborative communities, open source tools, and AI models, anyone can harness the power of safe, trustworthy AI for the next revolutionary idea, and that, you can print. (bright music fades)

The Future of AI is Open

Vox helps you understand our world.

Vox video is Bridgett Henwood, Cath Spangler, Adam Freelander, Joey Sendaydiego, Coleman Lowndes, Christophe Haubursin, Kim Mas, Laura Bult, Ed Vega, Rajaa Elidrissi, and with research support from Melissa Hirsch. 

Become a member today: https://www.vox.com/support-now

Check out our FAQ on permissions and licensing of our videos: https://www.voxmedia.com/pages/licensing

✉️ Write us: voxvideo@vox.com


The Future of AI is Open

스크립트