Science fiction films like Blade Runner and The Terminator offer analogies for understanding AI behaviors such as hallucinations and prompt engineering.
While code itself is deterministic with version control and infrastructure as code, the overall software process involves non-deterministic elements like user behavior and cloud reliability.
AI models are programs written by computers that infer rules from data, but their development is a complex, experimental process often left to specialized researchers.
Building a large language model from the ground up is a resource-intensive and unnecessary distraction, similar to building your own compiler just to write a blog post.
AI and LLMs represent a new computing paradigm, much like the smartphone, that enables novel user experiences by providing a conversational interface to existing tools and services.
Software developers should focus on integrating existing AI tools and applying their expertise in testing, deployment, and maintainability to build robust AI-powered applications.
32:26Krzysztof Cieślak
35:16Ankit Patel
25:17Scott Hanselman
28:27PJ Hagerty
47:35Laurie Voss, Rey Bango, Hannah Foxwell, Rizel Scarlett & Thomas Steiner
42:38Krzystof Czieslak
41:06Lars Gentsch
30:36Mackenzie Jackson
Jobs that call for the skills explored in this talk.
webLyzard
Vienna, Austria
Deyan7 GmbH & Co. KG