AI Workshop: Lessons learned

Event Details

We kicked off the workshops with a detailed breakdown of the current landscape of large language models, highlighting both their potential and present limitations. This involved an engaging discussion on pressing issues such as the limited context window of LLMs, the high costs and slow response times of GPT-4, and prevalent privacy concerns, among other topics. Our approach was both critical and solution-oriented, aiming to pave the way for advancements in the field.

Following the immersive introduction, attendees had the chance to get hands-on experience with leading AI tools in the format of a mini-competition. As the technical skills of the participants varied widely, we provided introductions to multiple different tools:

• Langchain – The prevailing framework for LLM development.

• Chainlit – A user-friendly UI framework that bypasses the need for frontend development.

• Langflow – A no-code tool that simplifies LangChain, albeit with certain limitations.

Using these tools, all of our participants showcased innovative use cases for AI like

• A PowerPoint karaoke generator built with Marp that generates an output including images and jokes in pptx

• An automated Gitlab merge request reviewer to streamline the code review process and enhance collaboration

• A meme generator that pushes the limits of current AI safety measures

• An interactive RPG Story in an invented world with dynamic image illustrations

• A study assistant to automatically generate relevant exam questions based on lecture slides or books

• A roasting chatbot that engages in witty banter and clever comebacks, providing an entertaining and light-hearted interaction.

After the workshop, it was time to unwind and reflect over slices of pizza. It was wonderful seeing the beaming faces, feeling the camaraderie, and sharing laughs as we looked back at the day’s accomplishments.

Lessons learned

Lesson 1: Open Dialogue Spurs Innovation

During my introductory talk on large language models (LLMs), we opened the floor to critical discussions, shedding light on both the potential and the current limitations of LLMs. During this discussion, multiple people raised completely novel concepts and ideas that went far beyond the content of the presentation. From this, we learned that fostering an environment of open dialogue can indeed spur innovative solutions and new approaches to existing challenges.

Lesson 2: Hands-On Experience is Invaluable

As participants got down to experimenting with tools like Langchain and Chainlit, it became evidently clear that there’s no substitute for hands-on experience. The small groups (3-5 people) allowed for a collaborative and intimate setting, encouraging everyone to learn by doing, which seemed to be a crucial catalyst in nurturing creativity.

Lesson 3: Give People Time!

We originally allocated around 90 minutes for the practical component of the event. Once people actually got into it though, it quickly became apparent that this would not be sufficient. Thus, we extended the time to allow all teams to accomplish the goals they set out to achieve.

Lesson 4: Unexpected Solutions are Often the Best Ones

I was both surprised and thrilled to see the ways participants leveraged AI to craft unexpected solutions. It was a stark reminder that sometimes the most remarkable ideas come from exploring uncharted waters and pushing beyond the conventional boundaries.

Lesson 5: End on a high note!

Transitioning the workshop into a casual pizza party not only allowed us to unwind but fostered a space of informal networking and connection. This fun conclusion reaffirmed my belief that ending on a high note enhances bonds and leaves everyone with a sense of camaraderie and anticipation for what’s to come

Your Turn to Learn with Us

If you are keen to be part of this learning and collaborative journey, I invite you to join us in our next workshop. Secure your spot by registering at the SelectCode Talent Pool.

And if you’re pondering over how AI can enhance your business, I’m just an email away. Connect with me at gereon.elvers@selectcode.de to explore the possibilities together.

Looking forward to learning and growing with you in our upcoming workshops!