From the course: The New AI Tech Stack: AI Literacy for Tech Leaders
Acquiring AI software
From the course: The New AI Tech Stack: AI Literacy for Tech Leaders
Acquiring AI software
- Since the introduction of ChatGPT in November, 2022, the world has raised to build AI solutions as an answer to business problems. At some point, there were 1,000 new AI applications released every month. That means there are tens of thousands of AI solutions you can choose from, and since the majority are relatively new companies, you need to know how to navigate through them and determine if there is a system that fits your needs. So let's take a look at what types of AI solutions there are out there. First, you have fully-trained systems that are ready to use and pre-trained AI solutions that expect you to use your own data to make the system speak your language. Then you have to choose between open source or proprietary systems. Open source is a system you don't need to pay for. Some of them are completely free for commercial use like Llama, and some are free for research and not for companies. For proprietary systems, you need to pay. Finally, you need to think about how the model is served. Some systems are fully downloadable, meaning that you can download them on your server or even your computer and fine-tune them as you please. Some systems are software as a service, so-called SaaS or MaaS, which is model as a service. A third option is where you use or download open-source or proprietary applications through the cloud. Okay, so now you know what types there are, but where to find them? I've compiled some resources for you in a downloadable handout, but here are some ideas to get you started. If you use cloud already, whether it's Azure, Google Cloud, or AWS, check what's there, and if any application fits your needs. There are also other clouds and deployment environments like Nvidia NIM or Anyscale. Also, I highly recommend Hugging Face. It's the most popular place to download LLMs and other AI software easily. Hugging Face has several leader boards where you can check the best LLMs, both open source and proprietary. As the next step, I usually check on a couple of popular websites where you can look for AI apps and buy or download them on the vendor sites. The first one is There's An AI For That, which offers over 10,000 useful solutions with an easy-to-use search engine and categorization. Or you can use Product Hunt or Simplified. They are very similar websites. There will be more and more platforms like these, including those that specialize in particular industries. Then I usually visit GitHub where you can find brilliant solutions, mostly open-sourced. But be careful and read the documentation as some solutions are still in beta. For generative AI, there's also the option of checking out OpenAI's GPTs and Microsoft Copilot, where companies and individuals upload their fine-tune version of chatbots and GenAI solutions. It's like the app store, but for AI. And finally, I use search engines, mostly to find proprietary solutions from larger suppliers. Then, based on all of this information and taking into consideration how much quality data I have available, I decide whether to build, buy, fine-tune, retrain, or use cloud for my AI solutions.
Practice while you learn with exercise files
Download the files the instructor uses to teach the course. Follow along and learn by watching, listening and practicing.