Ben Evans, a respected JVM technology architect and software expert, and Principal Engineer at RedHat, shares his insights into the evolution of artificial intelligence and observability in the current technological landscape. From the changing nature of AI to the challenges and opportunities of observability in containerized and cloud applications, we delve into the complexities facing the world of software.
According to Ben, Artificial Intelligence models face the following issues:
Sustainable Costs: Advanced AI solutions, such as language models, are extremely expensive to develop and maintain. Often, companies have not adequately considered the actual costs of operating these models, raising concerns about their long-term sustainability.
Limited Accuracy: Evans points out that current AI models do not always produce accurate and reliable results. He highlights a case in which models like ChatGPT experienced a 10% decrease in traffic due to the lack of accuracy in their responses.
Negative Feedback Cycles: The expert raises the possibility that AI models might worsen over time due to negative feedback cycles. In other words, if the data generated by the model is used for future training, it could introduce errors and reduce the overall quality of results.
Limits of Web Training: Evans questions how to maintain the separation between the training dataset and the online corpus, especially in models trained with Internet data. This poses challenges related to intellectual property and the possibility of the model generating text based on its own training.
Focus on Restricted Domains: Ben suggests that the focus of AI should shift towards highly domain-specific models instead of attempting to address the generalization of online text. He argues that using private and specific corporate data could lead to more accurate and useful responses.
In summary, Ben Evans emphasizes the need to address issues related to costs, accuracy, and negative feedback cycles in artificial intelligence, while exploring more focused and specific solutions in the field of observability.
Ben Evans is a JVM technology architect and a prominent figure in the realm of observability and software development. He has held notable roles in companies like New Relic and Red Hat Runtimes, and co-founded jClarity, a performance tool startup acquired by Microsoft. Ben is a Java Champion and has been a member of the Java Community Process Executive Committee for 6 years, contributing to defining new standards for the Java platform. Additionally, he is the author of several technical books and is recognized as a speaker and educator on topics such as the Java platform, system architecture, and observability.
Ben's talk at #GSAS23 will be called "The Why & What of Observability in 2023"
A brief summary of Ben's talk at GSAS 2023: As applications move to containers and migrate to the cloud, they become ever more complex, and it's increasingly important to monitor, analyze, and diagnose their behaviour. Observability is a new way of thinking about monitoring and understanding your applications. It’s supported by a growing range of open source tools and standards - part of the new wave of technologies that modern developers need to go fully Cloud Native. Join Ben Evans to get the basics of Observability, including the fundamentals of metrics, logs, and tracing. You'll learn how to achieve observability using OpenTelemetry, the emerging Open Standard for Observability, and we'll meet several other related OSS libraries and tools along the way.
The Global Software Architecture Summit (GSAS) is a 3-day event that aims to attract and connect software architecture experts from all over the world as well as all those interested in building working software to improve their skills, share knowledge, and connect.
The event features two days of talks by industry experts such as Mark Richards, Neal Ford, Nathaniel Schutta and one day of workshops. It is focused on software architecture topics as backend & frontend development, DDD, mobile development techniques, software architecture models & beyond.
The summit is held every year, so if you are interested in attending do not hesitate to buy your ticket. It is an excellent opportunity to give your career a boost and meet professionals in the industry.
EVENT ORGANIZER: APIUMHUB
Apiumhub is a software development hub that solves complex problems with high-quality solutions. Our purpose is to build scalable software to help our clients strengthen their businesses with powerful technology. Every year, we host this event in Barcelona to support the software architecture community and provide a space for the exchange of new and interesting ideas.
GSAS website: https://gsas.io/
Organizer site: https://apiumhub.com/