In this post I will discuss Artificial Intelligence and give you 7 tips to help you gain an understanding of AI.
I truly believe that AI will prove to be a game changer in the near future and will have a profound impact in all our lives.
Some of the impacts of AI and large language models will be detrimental and malign. But remarkable opportunities will also arise.
It is my firm belief that to ignore AI and Large language models would be a massive mistake.
And learning as much as you can to widen and deepen your understanding of AI makes a lot of sense.
Let’s take a look at 7 things I believe you need to understand:
- Learn machine English
Large learning models-ChatGPT, Gemini etc.-are not search engines. They do not store results of queries like Google search, or any other search engine. AI looks at the prompt and the context and generates, based on probability, the best outcome/result.
This is based on probability and proximity.
Takeaway: the prompt needs to be sharp, not vague. The AI large learning model is a guessing machine.
How to give good prompts: use the acronym AIM
A-actor-who is the machine acting as?
I-input-what is the input/context?
M-mission-what outcome are you looking for?
This 3 part structure can be used in all prompts. Structured prompts are vital.
- Pick one LLM and go deep
This allows you to build a relationship with it and use it as a sparring partner. I am going to focus on Gemini because I am very much embedded in the Google ecosystem as I use Google Drive, Gmail etc. for my solicitor’s practice.
ChatGPT is another option.
- Give context in your prompts.
Context is vitally important. Don’t be afraid to use assets such as files, data, documents to give the context.
Tell it what actions you are looking for it to take: search, scan, write?
The richer the context, the better the and more accurate the results.
- Debug your thinking
If you are not getting the right result it is probably your thinking that is the problem.Prompting is not typing. It is iterating.
Did you give it the right persona? the right context? the right goal/outcome?
You can ask the machine “what did you do?” “How did you arrive at that answer ?” “Show your reasoning”.
It will explain its chain. This is where the magic happens when you are using Large Language models.
Test, tweak, tune up the model.
- Steer to experts
Steer the model from mediocrity to expert sources. Do this with your prompts. List the sources, research papers etc. you want it to use.
- Verify what you get back
This is vitally important. Ask it about
- Its assumptions
- sources
- citations
- find counter evidence
- ask to audit
- cross model verification-you can use another model to test/check the answer you have been given
7. Develop taste in your output
-treat it like your sparring partner
-push it
-ask it to back its claims with evidence
-show your evidence
-state your thesis
Conclusion
Treat AI as your sparring partner or teammate.
Challenge it, ask it to stand over its conclusions/answers, ask it to cite sources, citations, experts, legislation etc.
Check out nanobanana for images, perplexity for sources, Sora 2 for video, ideogram