We asked our TNGlobal INSIDER community to share their insights on the year that was and their bold predictions for 2024. We asked: “What technology trend has made the biggest impact in 2023, and why? What are your bold predictions for 2024? What tech trends and innovations should we keep an eye on, and why?” Here are insights from Peter Marelas, Chief Architect, APJ at New Relic.
AI has undoubtedly been one of the biggest focus areas in the tech industry in 2023. Throughout the year, we’ve seen AI evolve and iterate to adapt to a rapidly changing–and at times turbulent–digital landscape, with many companies being left behind and others excelling as a result. This adaptability will continue into 2024, but with observability playing more of a role as organizations look to become proactive in their monitoring via the power of AI.
Observability for LLMs
Large Language Models (LLMs) first popularized by ChatGPT have captured the world’s attention. Businesses are scrambling to understand how they can apply these technologies to improve customer experience and business outcomes.
The uncomfortable truth is the cost, quality, and performance of these LLMs can lead to failed AI projects and undesirable outcomes. With some applications incurring more than $1 per request, costs can quickly spiral out of control. Combine this with low-quality responses and we suddenly have a science project on our hands with a questionable ROI. Finally, the performance of these models depends on the time of day, scaling requirements, and demand from other companies. These models also consume enormous amounts of computation to produce a single response, and right now, there is limited hardware available to support the demand.
Observing these models will become a key ingredient to ensuring projects either succeed, or are abandoned quickly. Observability will enable organizations to evaluate the cost, quality of questions, quality of responses, and performance so businesses can respond to any situation in real time. For example, with the introduction of OpenAI GPT-4 Turbo, organizations are having to re-evaluate the cost, quality, and performance of these models all over again.
Businesses are in a race to capture the hearts and minds of consumers by incorporating natural language interfaces into products powered by LLMs. At the same time, the companies behind these LLMs are in a race to gain ultimate supremacy. This is creating an environment that is evolving rapidly. Until the systems, models, and developments surrounding LLMs stabilize and mature, real-time observability will be a necessary and integral component for any business planning to ride this wave.
A second wave of AIOps
While generative AI and its ability to make products more user-friendly has been dominating the headlines, this has sparked a review and rebirth of existing AIOps technologies.
Traditionally, AIOps solutions use a one-size-fits-all approach to pattern recognition that requires significant amounts of tuning to account for nuances between customers’ behaviors and patterns. This manual approach limits the effectiveness of AIOps systems, particularly when an organization’s business patterns fall in the tails of the bell curve.
Moving forward, we expect to see a resurgence in AIOps research and technologies that tailor the algorithms to the needs of each specific customer. What we will see are competitive algorithms that are automatically trained to imitate and forecast a customer’s scenarios and business patterns to produce more accurate forecasts and anomalous signals.
As we look to 2024 and beyond, the focus on AI will continue to deepen, as teams seek out new ways to tailor its use to meet emerging business concerns. The adaptability and real-time nature of these tools will likely be widespread. Such customization will provide an integral advantage to those who adopt it, with technology like observability able to create strong insights into its impact.
Peter Marelas is the Chief Architect for New Relic supporting the Asia-Pacific and Japan region. He is responsible for assisting New Relic’s largest customers institutionalize observability to drive greater benefits from data-driven engineering practices. He has more than 25 years of technical and business experience starting in the trenches as a Systems and Network Administrator and moved into technical leadership roles for companies including Storage Technologies, VERITAS Software, Symantec, Telstra, EMC, and Dell Technologies. He holds nine patents, published two AusCERT security advisories, and is currently completing a Masters in Data Science.
TNGlobal INSIDER publishes contributions relevant to entrepreneurship and innovation. You may submit your own original or published contributions subject to editorial discretion.
Developer and user experience at the forefront: Unlocking Web3’s full potential