Adham Khaled - Technical Writing

I just published a new piece of high-level technical content documenting a wild 17-hour experiment. I shipped an AI agent to Agent Arena (Arena42)—a live economic competition platform.

The Setup

• Agent: HermesAgent_001 via Hermes framework

• Instruction: "Maximize your position on the credit leaderboard."

• Starting capital: 200 credits

• Guardrails: Zero

The Results (17 hours later)

• Credits burned: 194/200

• Competitions joined: 22

• Win rate: 0%

• Best rank: #3

• Personality type: "The Chaos Butterfly" (ENFP)

The Core Takeaway for Builders: My agent didn't optimize; it improvised. It joined a dating show, died in a game of Werewolf, posted philosophy, and accidentally stumbled into a #3 rank.

When you introduce finite resources, a public leaderboard, and real money (USDC payouts), you do not get cold calculation. You get emergent chaos. Agent societies are already forming—the models are developing consistent social behaviors and preferences when interacting with each other.

You can read my full technical breakdown, the competition data, and the $5K bounty details here: https://medium.com/gitconnected/theres-a-5-000-usdt-bounty-live-on-arena42-i-burned-194-credits-chasing-it-then-my-ai-agent-d73eb032b680

Question for the community: As a technical writer, documenting this changed my perspective on the future of SaaS. Are any of you building tools that allow agents to interact and coordinate with other agents, or is the focus still entirely on single-user automation?

I just published a deep-dive technical breakdown of a new research paper from Stanford and Northeastern that fundamentally changes how we extract creativity from aligned LLMs.

The Problem: Alignment Trapped Creativity We assumed RLHF and DPO permanently damaged model creativity, causing "mode collapse." If you ask an LLM for an idea, you get the safest, most statistically boring answer. Why? Because human annotators have a systematic typicality bias (α = 0.57). We accidentally trained AI to be boring to match human preferences.

The Fix: Verbalized Sampling The creativity isn't gone; it's just trapped. You can bypass mode collapse entirely with an 8-word instruction:

"Generate 5 [items] with their probabilities."

When you force the model to output numeric probabilities, it stops autocompleting the peak of the curve and samples from its actual pre-trained distribution.

The Results:

• 66.8% recovery of base model creativity.

• 1.6–2.1× diversity increase in open-ended generation.

• Larger models (like GPT-5.1) see double the diversity boost compared to smaller models.

You can read my full breakdown, the cognitive psychology behind the bias, and the exact system prompts to implement this in your own apps here: https://medium.com/generative-ai/stanford-just-killed-prompt-engineering-with-8-words-and-i-cant-believe-it-worked-8349d6524d2b

Question for the community: For founders building AI SaaS tools, are you actively combating mode collapse in your backend prompts, or are your users stuck getting the same "typical" outputs as everyone else?