Artificial Intelligence #87: New low-code data scientist course for domain experts and industry professionals who are non-developers

Seemingly out of nowhere - generative AI has become the technology of the year for 2022

Last year, I made a curious prediction: large language models (LLMs) will take off coupled with low-code development technologies

We at the #universityofoxford have been following this trend and it could be very disruptive in a good way because it could democratise the adoption of AI to domain experts (non-developers) to what I called the rise of the low-code data scientist??

We launched a course on Low-code data scientist: creating AI applications using low-code (online)

As it suggests, it's an opportunity for industry professionals and domain experts (non developers) to use low-code tools (Azure, AWS, GCP) to learn end to end development of AI applications.?

A unique feature of the course is the ability for creators of AI applications to collaborate with generative AI applications (such as GPT and chatGPT)?

The course is led by me and Ayse. We have been following generative AI technologies based on our collaboration with Pinckney Benedict and his team at SIU. We started exploring GPT-3? through the Open-AI APIs

This experience led me to believe even more about the combination of low-code and generative technologies working together

Why do we say this?

Think of it a bit like Client-server - stronger server = thinner client i.e more functionality goes to LLMs

To put this in context, I am old enough to know about Case tools ?and 4GL technologies like ABAP to appreciate that low-code is not a panacea. But low-code with generative models is very interesting. Not only do you get the benefits of low-code but you also get the benefits of generative models

That means

• Your starting point for development changes i.e. you start with a pre-built model
• All phases of the pipeline are affected? - code generation, design, data engineering, UX(chat interface) etc
• We democratise AI. Now, we bring empower domain experts to become data scientists
• By leveraging the cloud (In our case: Azure, AWS, GCP), we can create end to end pipelines in low code including MLOps

You can see this already with a combination of PowerBI and chatGPT for design

New skills will need to be developed such as developing prompting strategies for AI and understanding zero-shot, one-shot and few-shot learning

So, to conclude, unlike many experts in AI, I do not dismiss chatGPT/GPT-3. Instead, I see it as an opportunity to expand the AI talent pool base and develop new AI applications.?

?The course is here Low-code data scientist: creating AI applications using low-code (online)

?Themes covered by the course include:?

Machine learning and deep learning concepts?

• Building blocks of low-code ecosystems??
• Implementing the machine learning pipeline in low-code systems???
• Design and development of AI applications in the three low-code platforms: AWS, Azure, GCP (e.g. Microsoft Power platforms, Amazon SageMaker and Google Vertex AI platform)?
• Deploying low-code AI services in cloud native MLOps environments??
• Implementing generative technologies in low-code platforms such as chatGPT?
• UX and integration for low-code??
• Use of pre-built models and templates for creating AI applications?
• Comparing and contrasting capabilities of low-code and full-code AI solutions??
• Scaling low-code AI solutions?
• Security and access for low-code applications?
• Examples of low-code applications include: image recognition, business-based applications (e.g. invoice processing), sentiment analysis, time series, regression analysis?
• Creating and managing datasets for low-code applications: labelling data, dataset creation and annotation etc.?
• Workflow and process automation for low-code applications?
• Ethical AI and Responsible AI considerations for low-code and generative applications?

Themes covering the integration of low-code and generative AI applications include:??

• Introduction to large language models??
• Building AI applications using large language models?
• Understanding OpenAI, GPT and chatGPT?
• Introducing GPT-3 and the OpenAI API?
• Strategies for designing prompts?
• Understanding OpenAI models: Davinci, Babbage, Curie, and Ada??
• Working with the OpenAI Playground?
• Working with the OpenAI API?

If you are interested, please register your interest here or message me for any questions? Low-code data scientist: creating AI applications using low-code (online)