AI image generated by DALL-E with the prompt: “participating in a generative AI hackathon.”

Generative AI is taking the tech world by storm and has already developed to the point that a full stack ecosystem exists, allowing for the rapid prototyping and development of a new generation of applications and innovations.

New ideas and use cases of generative AI will be on display this Saturday at the AI Tinkerers Summer Hackathon in downtown Seattle.

“I think it’ll probably be the biggest Gen AI hackathon that we’ve seen in our community yet,” said Jon Turow, a partner at Madrona Venture Group, former Amazon Web Services product leader and one of the organizers of the event. “We’ve increased the size and yet we’re still massively oversubscribed.”

Generative AI is a powerful new approach that began when Google introduced a deep learning model called a transformer in 2017. The next year, OpenAI applied this approach to its first generative pre-trained transformer, or GPT. Subsequent versions of GPT and many other large language models (LLMs) have resulted in rapid advancement throughout the field.

That speed is being reflected in the development cycle as well, which has important implications for the pace of a hackathon. 

“Generative AI has implications for how fast you can move,” Turow said.

Popular models now include LLMs and multimodal models like GPT-4, ChatGPT, and PaLM 2, as well as image generators like Midjourney, DALL-E, and Stable Diffusion.

Though the AI foundational models are often seen as the workhorses of the field, the ecosystem has grown to include application frameworks that simplify the development process. Vector databases allow unstructured data to be vectorized and stored so it can be queried effectively. Training platforms allow developers to train their own language models. Prompt engineering, evaluation platforms, deployment services and many other tools further extend this evolving ecosystem.

Given how much is happening in the field, organizers hope to see all manner of different apps and solutions explored at the event. 

“It feels like a magic wand, so you’ve just got to jump in and tinker.”

“You’ve got a whole new tool in your toolkit now with the AI language models, and in generative AI in particular,” said Joe Heitzeberg, co-founder and CEO of Seattle startup Blueprint AI, and also one of the event organizers. “We don’t really know what’s even possible yet. It feels like a magic wand, so you’ve just got to jump in and tinker.”

Heitzeberg said events like this are major drivers of innovation, noting how giant tech companies including pet-sitting platform Rover and business communication company Twilio can trace their origins to hackathons and meetups.

Generative AI models can utilize a unimodal or a multimodal approach, which also expands the number of possibilities for innovation. A unimodal model uses one type of data — text, images, video, code, molecules, music and so on. Multimodal models use more than one data type in order to improve certain capabilities of the system. In a way, this is analogous to how we can make more informed decisions about our world when we use more than just one of our senses.

The advancement of generative AI has also led to concerns regarding its impact on areas like jobs, intellectual property and misinformation. As with so many past technologies, Heitzeberg believes the answer lies in diving in and discovering what the technology can do for us.

“When you make something better, faster, cheaper — even if it can be used for good and evil — it will generally be used for good,” he said. “And in doing so, it will enable new use cases you never thought of. Change is hard for people but ultimately, it nets out.”

The importance of discovering new capabilities is perhaps even more evident with an emergent technology like generative AI.

“The key inexhaustible resource of humanity is our creativity to find new important things to do and ways to do them,” Turow said. “When you add the kind of leverage generative AI offers to each of the minds that we have in the world, it’s really powerful.”

Given the level of interest for Saturday’s hackathon, organizers hope to be able to hold a bigger event in the fall, one that allows more people to participate. 

“Tinkering is important because it lets us plant this seed in everybody’s mind,” Turow said. “I’m sure that planting all these seeds will make the future happen faster and in a more positive way.”

The AI Tinkerers Summer Hackathon is being held June 10 from 8:30 a.m. to 8:00 p.m. at Create 33, the Seattle-based entrepreneurial center established by Madrona Venture Group.

The projects will be judged by a distinguished panel drawn from the AI industry, including Turow, Ivan Zhang (co-founder and CTO of Cohere), Vijaye Raji (founder & CEO of Statsig), Justin Uberti (co-founder and CTO, Fixie.ai), and others.

The event is sponsored by AWS Startups, Cohere and Madrona Venture Group with support from Pinecone, Weaviate and Blueprint AI.

Like what you're reading? Subscribe to GeekWire's free newsletters to catch every headline

Job Listings on GeekWork

Find more jobs on GeekWork. Employers, post a job here.