Home / News / Technology / OpenAI, Colin Jarvis: “Don’t Build for What’s Available Today, Build for the Capabilities That Are Coming”
Technology
3 min read

OpenAI, Colin Jarvis: “Don’t Build for What’s Available Today, Build for the Capabilities That Are Coming”

Last Updated June 12, 2024 12:02 PM
Samantha Dunn
Last Updated June 12, 2024 12:02 PM

Key Takeaways

  • Colin Jarvis opened up the AI Summit London 2024
  • OpenAI’s Chief Architect shared his insights into the future of AI in business.
  • He stressed the importance of innovating in ways that provide unique value, not just copying existing models.

Colin Jarvis, Chief Architect at OpenAI kicked off the AI Summit London with a keynote speech. His topic: “What’s Next for Language Models?”

Though OpenAI’s representative did not allow any follow-up questions, he shared insights into OpenAI’s vision of the future applications of language models, focusing on how AI businesses can differentiate themselves from others.

Multilingual Customer Service

During the keynote Jarvis spoke about the diverse use cases LLMs present, emphasizing their capacity to revolutionize customer service through multilingual capabilities.

“You can write an application for an English-speaking market and it can be applicable in other countries,” Jarvis noted, touching on the expansion potential that language models offer.

OpenAI Colin Jarvis at the AI Summit.
Photo Credit: Samantha Dunn, CCN

Demonstrating how close the top ten chatbots are now, in terms of textual intelligence, Jarvis highlighted how high-performing AI providers are now very close to each other. Describing this as an “AI Arms Race”, Jarvis noted that organizations will need to differentiate themselves in key areas if they hope to compete.

AI’s Four Areas of Development

Addressing where businesses can find tangible value, Jarvis pointed to four key areas for development:

  1. Textual Intelligence: The ability to generate, understand, and manipulate text in increasingly sophisticated ways.
  2. Economical Models: Advancements that make models cheaper and faster, reducing the cost of intelligence per dollar.
  3. Increased Customization: As models become more customizable, they can be tailored to specific business needs, albeit with new safety and security challenges.
  4. Multi-Modal Integration: Future models will likely incorporate multiple modalities—text, image, and beyond—within a single framework, broadening their applicability and functionality.

Building for the Future, Not the Present

One of Jarvis’s key messages was the need for businesses to think ahead. “Don’t build for what’s available today,” he urged. “Build for the capabilities that are coming.”

Increasing multimodality specialisation
Photo Credit: Samantha Dunn, CCN

He emphasized that while these powerful models are accessible to everyone, the real advantage comes from how they are utilized. The user experience, the data used, and the degree of customization will distinguish successful implementations from the rest.

Strategic Advice for Businesses

In his closing remarks, Jarvis offered strategic advice for businesses looking to leverage AI. He stressed the importance of innovating in ways that provide unique value, not just wrapping around existing models.

“Everyone has access to these models,” Jarvis explained. “The user experience, the data you bring, and how you customize—this is how to differentiate and build something unique. If you just build a wrap around these models, you’re no different to your competitors.”

Was this Article helpful? Yes No