- Published on
Where should you start learning about AI?
With the prominence of AI continuing to rise blogs, industry newsletters and LinkedIn are all packed with new AI and LLM launches, investments and products every single day. For a lot of people it is probably starting to feel like you're being left behind in terms of knowledge, and finding a starting point to introduce yourself to the world of AI is getting more and more tricky.
Friends, colleagues and users regularly ask us for our tips on places to start on the journey so Superface CTO, Z, pulled together a list of resources that we've been sharing with them. We figured why not put it out there for everyone, so here it is; our tips for Product Managers, Testers, CIOs, CTOs and others.
Experiment with ChatGPT
It sounds obvious but as a starting point digging into ChatGPT won't steer you too far wrong. It's one thing to know about it, it's another to actively use it for something more than an alternative to Google. What kind of experimentation should you try? See below.
Try Superface GPT with ChatGPT
One of the ways you can experiment with ChatGPT in a way that supports your day-to-day, with very minimal additional setup, is to use this custom GPT that allows you to interact with your Google Sheets, Calendar, Mail or SaaS apps like Slack or Jira.
Read the Prompting Guide
Chat interface writers block is certainly a reality, so understanding the how, why and what of prompting LLMs can help unblock both productivity and the quality of the results you get from any models you work with. We recommend starting with the Basics of Prompting first.
Go beyond OpenAI
There are other lots of other popular models and the first one you work with might not be best suited to your use case. Using the Prompting Guide, try the same prompts with multiple models to understand how they handle what you're asking.
Checkout HuggingFace
If you want more models. All of them. HuggingFace is where the machine learning community collaborates on models, datasets, and applications. They also have their own HuggingChat chat interface for most of the models listed above as well as the ability to create and try out different Assistants built on those models.
Learn about RAG
Retrieval Augmented Generation is an approach that can be used on top of generic language models (most of the popular LLMs fall into this category) to fine tune their abilities when it comes to working with specific documents or datasets. RAG is particularly useful in a business context because it enables more factual consistency, improves the reliability of responses, and helps to mitigate the problem of model "hallucination".
Learn about LangChain
However, understand that it is not always the right framework for the job
Build something with LangChain
If you feel like writing some code, then a Jupyter notebook, or Google Collab can get you started without the need for any additional tools. You can use LangChain directly from either of these to interact with models, vector stores and custom libraries for a wide variety of tasks.
Experiment with AutoGPT
For the more adventurous, and non code averse, AutoGPT is a great place to start if you're looking to dig deeper into the world of "autonomous agents". AutoGPT helps to automate the back-and-forth interactions with chat-driven models like OpenAI. You start by describing what you're looking to achieve and AutoGPT will handle the necessary communication to make it happen.
Keep up with the newsletters and podcasts
There's so much news in this space that keeping up with it can actually be left to the professionals who produce these newsletters and podcasts, so you don't have to.
- AI Weekly
- The Algorithm from MIT
- Steampunk AI newsletter
- AI in Business podcast
- Practical AI podcast
- This Week In Machine Learning podcast
There are many, many more to find so dig around to see if there are any that dive deep on specific niches that you're interested in.