Substack AI Image Generator; Prompt > AIaaS: The Power of Prompts
Context & Thanks
Thanks to Michael Spencer (MS) and Melanie Mitchell (MM) for starting the conversation; here’s the context:
Please note, MM posted:
“Chain of Thought” Reasoning in LLMs
In the last few years, there has been a deluge of papers making claims for reasoning abilities in LLMs (Huang & Chang give one recent survey). One of the most influential such papers (by Wei et al. from Google Research) proposed that so-called “Chain of Thought” (CoT) prompting elicits sophisticated reasoning abilities in these models. A CoT prompt gives one or more examples of a problem and the reasoning steps needed to solve it, and then poses a new problem. The LLM will then output text that follows a similar set of reasoning steps before outputting an answer. This figure from the paper by Wei et al. gives an example:
Wei et al. tested the CoT prompting method on datasets with math word problems, commonsense reasoning problems, and other domains. In many cases, CoT prompting dramatically improved the performance of LLMs on these datasets.
Two downsides of CoT prompting are (1) it takes some effort on the part of the human prompter to construct such prompts; and (2) often, being able to construct a CoT example to include in the prompt requires the prompter to already know how to solve the problem being given to the LLM!
A subsequent paper proposed that these problems can be avoided by leaving out the CoT example, and just posing the problem to an LLM with the added phrase “Let’s think step by step.” Here’s an example from that paper:
Amazingly, this simple addition to a prompt—termed “zero-shot CoT prompting”— substantially improves LLM’s performance on several reasoning benchmarks as compared with prompts without the addition of “Let’s think step by step.”
Why does CoT prompting—whether with an example or with “Let’s think step by step”—have this dramatic effect on LLMs? Interestingly, while there are many hypotheses, this question has not been definitively answered.
A key quote from MM:
Amazingly, this simple addition to a prompt—termed “zero-shot CoT prompting”— substantially improves LLM’s performance on several reasoning benchmarks as compared with prompts without the addition of “Let’s think step by step.”
Please note the power of prompts as it relates to AI. Seemingly simple changes in the input (i.e., “conversation”) with AI changes everything.
The power of prompts may be one of the single biggest lessons to learn as it relates to AI, especially if the goal is to monetize AI-as-a-Service (AIaaS).
Learning how to prompt (i.e. chat, converse, talk, etc.) AI isn’t as easy as it may initially seem, like other languages, it takes time and patience to learn.
The Opportunity
Let’s learn together through observation and use how humans (i.e. Natural Intelligence (NI)) are prompting (i.e. conversing) with AI bots powered by GPT-4.
AI is here. AI is like the internet, it’s not if but when and where we’ll encounter it; it’s also about NI (i.e. definition here) controlling AI vs. AI controlling NI
If we’ve used Amazon’s Alexa, Apple’s Siri, or even auto-correct when texting or typing, say hello to AI, because we’re using it
Since AI is here and we, and even our parents or grandparents, are using AI, wouldn’t it be wise to start to understand it better?
AI is all about the power of prompts (i.e. learn more AI vocabulary here); AI prompts are essentially a language in itself
Why not learn about conversing and controlling AI through prompts via the power-packed duo of observation and use?
Alex Chompff (AC) the creator of MasterVerse.AI believes in learning by observation and use so much that AC is giving un-throttled access to GPT-4 for free for now.
Please note, GPT-4 is neither free nor un-throttledif accessed directly (see here and here). To be answered by AI bots powered by GPT-4 costs money.
In addition to the cost of each answer by AI bots powered by GPT-4, there’s also the time, money, and energy costs of hosting and maintaining the services.
AC is funding MasterVerse.AI based on a mission to serve others. You may benefit from this by joining the MasterVerse.AI Slack Community.
If Slack isn’t your preferred platform, you may also learn about the power of AI prompts by subscribing to the MasterVerse.AI Substack via the button below.
Both free and paid MasterVerse.AI subscriptions offer:
You see the questions others are asking AI bots powered by GPT-4
This gives valuable insight; other platforms may give access to GPT-4, but users usually only see their own questions, not others
You see the answers AI bots powered by GPT-4 give in reply
Yet again more valuable insight, as you see what others see, and what others are paying for; who’s paying? Currently, MasterVerse.AI.
The free subscription lets you read posts within the past month.
The paid subscription lets you read posts without a time-limit. At the time of typing MasterVerse.AI Substack already has a library of 187 published posts ready to read.
i.e. Humans and Machines living and working together
Conclusion
AI is here. It’s really only a question of how much we realize and accept it. Accepting reality is where adapting begins. Let’s learn and master AI together.
Alex Chompff is generously funding two opportunities to adapt to AI by learning how to powerfully and successfully prompt AI:
Option 2: MasterVerse.AI Slack Community is a diverse group of NI (i.e., humans) working with a diverse group of Artificial Intelligence (AI) bots powered by GPT-4.
