All Transform 2021 sessions are available on demand now. Look now.
Executives should prioritize whether they want to experiment with cutting-edge technologies or wait to see results from other implementations first, Airbnb CTO Vanja Josifovski said in a conversation with VentureBeat Founder and CEO Matt Marshall at VentureBeat Transform 2021 Virtual Conference. Most companies, even the largest, have limited resources, so they have to decide which technologies to invest in and which to wait.
Typically, the decision is to use cutting edge technologies in critical areas and to avoid experimental or emerging technologies in all other areas, Josifovsky said.
“This is one of the hardest parts of my job because I want to hire the best and smartest people, but then I want to channel that ability into the areas that will have a business impact,” said Josifovsky. “In some cases, [we] refrain from using state of the art until we believe we will collect the return.
Josifovsky and Marshall discussed some of the innovative trends in artificial intelligence (AI). “If we look at what is happening today, there are some amazing technologies to come,” said Josifovsky, such as graphic neural networks, transformer models and language models.
Graphic neural networks
Graphic neural networks will be a major trend in 2021, Josifovsky predicted. At its core, the deep learning paradigm is a different way of structuring data, like images, and sequencing data, like text. However, the use of the data and the structure required to operate the model can be rigid. Graphic neural networks, on the other hand, allow a more flexible architecture because the data defines the architecture of the model.
“Graphical neural networks are a next iteration that allows us to use a lot more data in the deep learning framework in a much more natural way,” said Josifovski. “I think they’ll open up a whole new area, where you can apply the deep learning paradigm much more easily to a whole different set of data.”
Although this is “an incredible technological achievement,” it may be too early to work with large language models, Josifovsky said. Being able to scale these models is a relatively new concept, but the challenge is to find the data to train the model. However, he added that using models in the production process requires predictability. There have been good examples of using templates to generate text and web pages. This is a good place to use the models because they are not “mission essential,” said Josifovsky. On the other hand, this type of work will not fit well initially in machines like self-driving cars.
While language models do not currently work with chatbots, Josifovsky believes they will in the future.
In the early years, academia was the center of innovation and research for AI, with large companies developing proprietary technologies, Josifovski said. Over time, waves of innovation in AI have come from big companies, like Google, Amazon, Microsoft, and Facebook. As many technologies become commonplace, Josifovsky predicts another shift, this time to smaller independent companies. In areas like cloud infrastructure storage and management, well-endowed companies will provide the infrastructure for smaller players to develop AI.
“The center of gravity will shift slightly from large companies to small independent companies,” said Josifovsky. “We will see a full ecosystem of companies that [have] developed today and will shape the future.
VentureBeat’s mission is to be a digital public place for technical decision-makers to learn about transformative technology and conduct transactions. Our site provides essential information on data technologies and strategies to guide you in managing your organizations. We invite you to become a member of our community, to access:
- up-to-date information on the topics that interest you
- our newsletters
- Closed thought leader content and discounted access to our popular events, such as Transform 2021: Learn more
- networking features, and more