1. From Rendered to Generated: The Future of Visual Media
Prem Akkaraju, CEO of Stability AI, envisions a future where generative models transform Hollywood and beyond. Working previously on Avatar 2, he noted that:
- Some scenes took 6,000+ hours to render a single frame
- Generative tools can now compress that into minutes
- The future of content is 10x–20x more media, with diverse time formats (e.g., a 5-minute film just for your morning routine)
But Akkaraju emphasizes that creativity must remain human-led:
“The creative process starts with people. AI agents are tools, not auteurs.”
Instead of eliminating actors, directors will use AI to enhance performance, not replace it.
2. The New Modality: Beyond Language and Vision
Richard Socher, CEO & Founder of you.com and Co-Founder & Managing Director at AIX Ventures, explains that the future lies in multimodal intelligence:
- Text, images, voice, video, and code are all merging
- The next frontier? Biology
Socher reveals that large language models can now generate novel proteins — potentially transformative for medicine.
“You can ask an AI to write a haiku — or a protein that only binds to a cancer cell.”
This opens the door to a new kind of creativity: biological invention.
3. Intelligence Without Billions of Dollars
Dr. Kai-Fu Lee, Chairman & CEO of Sinovation Ventures and CEO of 01.AI, is proving that world-class models can be trained without elite access or massive spend.
- His team trained a top-tier LLM with just $3 million
- Inference cost: $0.10 per million tokens, vs $4.40+ for others
- Ranked #3 globally among foundation model developers, behind OpenAI and Google
“Necessity is the mother of innovation. Constraints fuel creativity — especially in China.”
Lee believes that with the right engineering focus, global accessibility is possible — even without Silicon Valley-level capital.
4. Self-Training and Infinite Productivity
AI is transforming productivity across fields — but its impact depends on digitisation. Socher explains:
“In simulatable environments like programming, AI can iterate infinitely. In messy real-world tasks like plumbing, it still needs human guidance.”
AI’s next phase involves training agents through natural instruction — reshaping how marketers, analysts, and creators spend time.
Real-world example: A cybersecurity team automated 20 hours of work per week just by describing a repetitive task to an agent.
5. The Coming AI Divide: Who Will Be Left Behind?
The panelists agree: companies will divide into two categories by 2030:
- Those who fully integrate AI
- And those left behind
“At the end of this decade, you’re either using AI — or you’re out of business.” – Dr. Kai-Fu Lee
6. Should the Next Generation Learn to Code?
Prem Akkaraju: “Don’t waste your time learning to code. The new language is English.”
Richard Socher: “Disagree. Programming helps you understand and build the future.”
Dr. Kai-Fu Lee: “Follow your passion — become either a world-class builder or a skilled AI manager.”
Final Thought: The Jevons Paradox of Intelligence
Efficiency increases usage. AI will make intelligence cheaper — and that means we’ll use it everywhere.
At Anjin Digital, we’re building for that future: helping brands and teams orchestrate AI workflows that elevate human creativity, decision-making, and scale.
Because the future isn’t AI vs. humans.
It’s humans with AI vs. humans without.
