Over a million developers have joined DZone.

AI/ML 2018 Surprises and 2019 Predictions

DZone's Guide to

AI/ML 2018 Surprises and 2019 Predictions

More real-world applications of AI/ML providing business value.

· AI Zone ·
Free Resource

Bias comes in a variety of forms, all of them potentially damaging to the efficacy of your ML algorithm. Read how Alegion's Chief Data Scientist discusses the source of most headlines about AI failures here.

Given the velocity with which technology is changing, we thought it would be interesting to ask IT professionals to share their thoughts on the biggest surprises in 2018 and their predictions for 2019. Here's what they told us about Artificial Intelligence (AI), Machine Learning (ML), and other iterations of data science:

Jayant Lakshmikanthan, CEO and Founder, CLARA analytics

2018 surprise — I have been surprised at the rapid pace AI has been applied to the InsureTech industry. But more so, I am impressed with how quickly commercial insurance carriers are realizing the need to start using AI in some part of their workflow such as underwriting, claims operations, and customer service to maintain a competitive edge.

2019 prediction - In the B2B AI space, there is going to be an increasing focus on hard dollar savings and value. The theoretical value proposition of AI is well accepted. But what companies will expect of their AI technology providers in the coming years is the realization of that value.

Scott Parker, Director of Product Marketing, Sinequa 

While there has been so much hype around ML and AI, the biggest surprise from 2018 is how much of it is still being tested in the lab. There is an abundance of marketing around the related technologies, but, with a few exceptions, AI projects remain in the lab waiting for compelling use cases to come along.

For 2019, ML and AI will finally find their way out of the lab and into existing applications one way or another. For the most part, people won’t even know it’s there because it will be embedded in a seamless fashion.

Minkyung Kang, Data Scientist, Aquicore

End-to-end machine learning services such as Databricks MLflow and Amazon SageMaker are making ML workflow much easier and simpler. Data scientists and developers are enabled to build, train, and manage ML models in one place and move the models to the production at scale without worrying too much about pipeline and architecture.

Predictions: Connecting and integrating different steps and processes in ML workflow will be further improved and streamlined and allow many startups and enterprises to move quickly with their ML applications with fewer resources. This will expand further to the management of the entire lifecycle of ML, including data collection and management.

Cormac Driver, Head of Product Engineering, Temboo 

We were surprised and delighted to learn about the unexpected ways in which artificially evolving organisms will adapt to solve a wide range of problems. For example, in one experiment, robots learned that small rounding errors in the math that calculated forces meant that they got a tiny bit of extra energy with motion. They learned to twitch rapidly, generating lots of free energy that they could harness. The programmer noticed the problem when the robots started swimming extraordinarily fast.

We predict the hype around AI will diminish, while real-world deployments will continue to increase. The most notable breakthroughs will come from fundamentally rethinking how we approach computing at the most basic level (see Facebook's recent reimagining of Floating Points to achieve a 70% increase in model training).

Peter Wang, Co-founder and CTO, Anaconda

Surprises in 2018

The Github and Red Hat acquisitions. Microsoft buying access to the developer community, and IBM buying software infrastructure for hybrid cloud future.

The Cloudera and Hortonworks merger was also a surprise. It marks the end of the Hadoop "big data" hype cycle, and makes it clear that future growth in analytics and ML must target heterogenous storage architectures.

Predictions for 2019

The industry will start forming standards and best practices around corporate data science. It's just too much of a wild west right now.

"Data Science" as a field will fragment into a few sub-specialties, including data engineering, advanced statistical inference, and storyteller/explainer.

As we learn more about the depths to which authoritarian nations use AI and a sensor-rich world to perfect the surveillance state, it will fuel momentum behind more data privacy legislation in the West.

Your machine learning project needs enormous amounts of training data to get to a production-ready confidence level. Get a checklist approach to assembling the combination of technology, workforce and project management skills you’ll need to prepare your own training data.

artificial intelligence ,machine learning ,ai ,ml ,deep learning

Opinions expressed by DZone contributors are their own.

{{ parent.title || parent.header.title}}

{{ parent.tldr }}

{{ parent.urlSource.name }}