AI Improves Performance via Human-like Prompts
In a research paper called "Large Language Models as Optimizers," Google DeepMind team developed a technique to improve math ability in AI language models like ChatGPT. How'd they do this???
By simply improving their prompting (the written instructions that tell the AI model what to do). They found that using human-style encouragement improved math skills dramatically. Which matches what prior research has shown.
Rather than relying on formal mathematical definitions to perform this task, the team used "meta-prompts" to set the stage for the optimization process. The LLM then generates candidate solutions based on the problem’s description and previous solutions, and it tests them by assigning each a quality score.
Phrases like "let's think step by step" prompted each AI model to produce more accurate results when tested against math problem data sets. (This technique became widely known in May 2022 thanks to a now-famous paper titled "Large Language Models are Zero-Shot Reasoners.")
Hockey Coach Connection
The connection with hockey coaches is obvious... when a player is struggling or being prompted by a coach, coaches should ensure players do not rush their thinking.
Just like researchers found that "take a deep breath and work on this problem step by step" was an effective phrase, so does a similar prompt fro a coach to a player.
- "Take a deep breath and work on this problem step by step" achieved the top accuracy score of 80.2 percent in tests.
- The classic "Let’s think step by step" prompt scored 71.8 percent accuracy.
Next time you see a player rushing through a situation... give them a simlpe drop to "take a deep breath and work on this step by step"