The Tech Conference That Wasn't Selling Anything: My Key Takeaways from Flower City AI

1. Community and 10-Year Predictions 🔮

The opening keynote, delivered by event founder and organizer Max Irwin (Max Irwin), reflected on the community's evolution: from Year 1 (basic education) to the current Year 3 theme of societal impact. Max emphasized the event thrives on community inertia, not advertising, a point proven by how I received my ticket from my friend N. Harris.

Max Irwin's 10-Year Forecast:

• Economic Shift: A huge resurgence in the trades and domestic manufacturing as younger generations pivot away from desk work.
• Ad Tech Struggles: The digital advertising economy is struggling, with search summaries causing click-through rates on ads and search results to drop by as much as 50%.
• New Services: Expect fully personalized education, AI-powered personal advocates, and new digital therapies.
• Legal Battles: The New York Times vs. OpenAI lawsuit may last 20 years.

2. Generative AI in Healthcare: The In-House Advantage 🩺

Carly Hochreiter from URMC shared their strategy for developing generative AI tools in-house, highlighting cost and customization benefits over vendor solutions.

Pre Chart Pro Tool & Pilot Results:

The tool targets pre-charting, where providers spend ~1/3 of their EHR time. It uses 9 clinical data streams and a hybrid LLM approach (GPT-4/4O, temperature=0).

Usability:       81%   # Rated easy to navigate
Helpfulness:     63%   # Highest benefit for complex patients
Time Savings:    63%   # Reported reduction in EHR time
Risk Reduction:  50%   # Reduced risk of missing information

3. Innovation and Infrastructure: The Rochester AI Hub Initiative 🚀

This talk provided a concrete local mandate: the Rochester AI Hub Initiative, aiming to make Rochester a top 5 global AI hub.

Total Funding Target:       $100M
Local Companies Involved:    80+
National Partners:           12
Infrastructure Goal:         Top 5 Global AI Hub
Focus Areas:                 Talent Pipeline, GPU Access, Research, Startups

4. The Data on Sustainable AI Usage ♻️

This crucial session on the environmental impact of AI was delivered by Madhura Anand (Madhura Anand), a data professional with the City of Rochester focused on policy and sustainability.

The Problem: Inference is the Energy Monster

Inference Dominates:  70% to 80% of model's total energy consumption
Energy Scale:         ChatGPT-4 model for one year needs ~441,000 megawatts
                      (enough power for 35,000 homes)
Inefficiency:         Response length biggest factor (85% correlation)
                      Large models waste 10-30x more energy for simple tasks

Solutions: Individual and Systemic

• Individual Action: Set max token limits (e.g., 200-300 tokens), choose smaller models, and stop prompts immediately.
• Systemic Change: Hardware fixes needed for GPU chips wasting 40% to 60% energy. Policy must require standardized, auditable disclosure of energy, carbon, and water usage.

Conclusion

The Flower City AI event successfully framed the technology not just as a fascinating tool for personal and societal advancement, but as a critical responsibility.

I realized this was a true community effort—unlike many conferences where talks revolve around a developer advocate pitching a product to hit team MRRs or KPIs, this event was simply about people, both in and out of tech, genuinely curious about the future of artificial intelligence.

We stand at a crossroads where we must navigate the "fascinating and terrifying" future by actively participating in its design.