Progress in generative AI

Generative AI Concepts

Daniel Tedesco

Data Lead, Google

A collaborative effort

Universities
Governments and civic organizations
Open-source communities
Startups and large companies

Universities

New research, such as invention of GANs
New researchers
Partnerships with other sectors

Homepage of the University of Montreal's AI center

¹ University of Montreal website

Governments and civic organizations

Governments

Establish and enforce regulatory environment
Fund basic research

Civic institutions

Provide independent analyses
Resources, such as datasets
Project funding

Image of conference exhibition booths with a banner overhead that reads "DARPA Demo Day"

¹ National Defense Magazine, https://www.nationaldefensemagazine.org/articles/2016/5/11/darpa-shows-off-technology-at-demo-day

Open-source communities

Provide open access to tools and models
Lower barriers to entry, experimentation, and sharing
Are difficult to sustain and maintain quality
Raise risks of misuse

Some generative AI Open-source projects

Collection of open source generative AI tool and model logos

Startups and large companies

Seek competitive advantage
Bring generative AI to broad adoption
Showcase advances to attract talent and impress investors

Large companies additionally:

Fund research, such as introduction of transformers by Google
Acquire startups
Offer hardware and cloud resources

The openness challenge

Pros

Developer support
Talent attraction
Broader feedback

Cons

Lose competitive advantage
Risk liability for misuse

The boundaries of generative AI development

Accelerators

Decreasing hardware costs
Research developments
Competitive and geopolitical pressures

Decelerators

Technological limits
Overbearing regulation
Closed ecosystems
Limited economic resources

Let's practice!

Generative AI Concepts

Preparing Video For Download...