Digital business depends on organizations’ ability to turn more data into more useful work. The principal enablers in that regard are enterprise investments in advanced analytics, artificial intelligence (AI), deep learning (DL), and machine learning (ML).
On Friday, December 15, Wikibon held our annual webcast in which we predicted how the data analytics market, platforms, tools, and practices are likely to evolve in the year to come. From that webcast, here are principal predictions from George Gilbert, Neil Raden, and myself:
- AI will become the backbone of high-performance IT operations and application management environments. ML is essential for continuous monitoring, optimization, diagnostics, and remediation of today’s increasingly complex distributed computing environments. Gilbert predicted that, in 2018, IT operations and application performance management will become ML’s first high-volume, horizontal application.
- Data-driven auto-programming will become a centerpiece of enterprise application development. I predicted that, by year-end 2018, the latest auto-programming techniques – including ML and robotic process automation (RPA)–will be incorporated into the top tier cloud-focused integrated development environments. What ML and RPA have in common is that they employ a domain-driven code-generation framework and use data-driven training of code-builds to boost their effectiveness.
- Data science productivity will rapidly improve, enabling more work to be accomplished with fewer data scientists. Raden predicted that, in 2018, the growing availability of tools for automation of data wrangling and modeling will enable data scientists to keep pace with growing demand for their services. This trend may even, with the democratization of this field, lead to a decline in demand for traditional data scientists.
- Analytics applications will become showcase microservices in the use of functional programming over serverless clouds. Functional programming addresses the core features of mainstream cloud microservices, including many new ML, DL, and AI applications. I predicted that, by year-end 2018, more than 50 percent of new microservices deployed in public cloud will be deployed in serverless environments. But due to the embryonic adoption of serverless, on-premises platforms, fewer than 10 percent of new functional code-builds in private clouds will use functional code.
- AI microservices will increasingly move to edge devices for autonomous operations, thanks to the growing use of reinforcement learning. I predicted that, by year-end 2018, more than 25 percent of enterprise AI application-development projects will involve autonomous edge devices and that, by that time, more than 50 percent of enterprise AI developers will have gained familiarity with reinforcement learning tools and techniques.
- Packaged, pre-trained machine learning models will become key to enterprise application strategies. Enterprises continue to face a shortage of data scientists to build, train, deploy, and manage AI and other data-driven models in production applications. Gilbert predicted that, in 2018, IT professionals will be less likely to build machine learning models from scratch. Instead, organizations will increasingly use and customize the API-accessible pre-trained models within packaged apps and cloud services.
- Vendors of open-source big-data software solutions will consolidate through mergers and acquisitions. Gilbert predicted that, in 2018, these consolidations will be driven both by investors’ increasing demand for profitability over sales growth and by the need of niche vendors to build enterprise salesforces to capitalize on their “land-and-expand” go-to-market strategies.
- Data movement to and from edge devices will become the predominant cost element in IoT operations. Raden predicted that, in 2018, approaches for reducing IoT data transmissions and compacting AI and other edge-deployed algorithms will gain popularity.
Please tune in here to watch the full webcast with slides.
