Above: This is what AI thinks the future of AI looks like. Image created with Midjourney.
We recently hosted a lunch and learn with leaders from Google Gemini and Open AI. Here are our 6 key takeaways for the future of AI.
Creature and 25madison hosted an insightful Lunch & Learn featuring two prominent voices in the world of artificial intelligence: Liam Bolling, a Former Product Manager at Google Gemini and an alum of the 25m Founders Club, and Ilan Bigio, an engineer at OpenAI. Their engaging discussion spanned a variety of critical topics, from the evolving landscape of AI technology to the pressing ethical questions it raises.
Here are some key learnings from our conversation:
- The AI revolution is a huge jump forward in humanity’s evolution
"We are living in a truly extraordinary time in human history. If you're working on a startup right now, you're at the forefront of a once-in-a-lifetime opportunity. For the first time ever, the cost of intelligence is plummeting, opening the door to create some truly revolutionary companies." - Liam Bolling
As a transformation studio, this electrifies us. When we asked about the rapid pace of AI advancements, Liam’s response was clear: "Innovation is unstoppable, and AI is only accelerating. There’s no hitting the brakes now."
- It’s a great time to build a business
Liam highlighted the unprecedented opportunities available in the AI space right now. He compared this moment to the launch of the first iPhone or the rise of the Internet, suggesting it’s even more revolutionary. "Intelligence is becoming cheaper and more accessible," he said, encouraging startups to seize the moment and "build some really cool stuff." - The businesses that leverage AI in unique ways are the ones that will win
When asked what kind of startups in the AI space will succeed, Liam said that companies that leverage AI in unique ways will succeed vs. those that are trying to improve upon the core functionality of the large language models themselves. For example, using AI to improve customer service in the hospitality industry—such as employing AI to handle customer support—represents an opportunity for startups that large companies like OpenAI and Google Gemini are unlikely to pursue themselves. OpenAI and Google Gemini are likely to continue making the core LLM continually better and more sophisticated, thus startups going after this space may be fighting a tough battle. - How will the models keep getting better?
The team mentioned that the major large language models have consumed much of the open-source and publicly available text-based data to train the current models. We were curious how the models will continue to improve if there is no more public data to ingest. Liam had a hot take when it came to User Generated Content and AI: we’re just scratching the surface! With so much video and audio content generated daily, there’s a wealth of data waiting to be harnessed. - Don’t rule out synthetic data
Ilan highlighted the exciting potential of synthetic data as a way forward. Synthetic data is computer-generated information designed to improve AI models, protect sensitive data, and mitigate bias. This method allows us to generate data that can train AI models effectively, even when real-world data is scarce. He reassured us that while many raise alarm bells about synthetic data, it’s actually proving to be a powerful tool for innovation. - AI still has a lot to learn from humans
Liam and Ilan emphasized the impressive efficiency with which humans learn, often drawing from minimal experiences to make complex decisions. Liam highlighted that, unlike AI models that can require vast datasets and significant computational resources to learn effectively, humans are capable of adapting and generalizing from just a few examples. This adaptability allows us to navigate new situations with ease, a trait that AI is still striving to emulate.
Ilan reinforced this idea by noting that the ultimate goal in AI development is to create models that replicate this human-like flexibility and efficiency. "We know it can be done," he said, expressing optimism that researchers can design algorithms that learn from smaller amounts of data, much like how humans can derive insights from everyday experiences. This would not only make AI more efficient but also more accessible, as it could operate without the extensive infrastructure typically associated with large-scale AI projects.
Stay tuned
We hope you enjoyed this recap of our Lunch & Learn with AI experts Liam Bolling and Ilan Bigio! Stay tuned for more discussions as we connect with other industry leaders to explore the exciting worlds of AI, technology, branding, marketing, user experience, and more.
