View all newsletters
Receive our newsletter - data, insights and analysis delivered to you
  1. Technology
  2. AI and automation
March 27, 2019updated 28 Jun 2022 7:41am

Automotive AI Hits a Speedbump: Pilot Projects Shrink, Deployment Falls

"A missed opportunity"

By CBR Staff Writer

Just 10 percent of major automotive companies are deploying Artificial Intelligence (AI) according to a major new study by professional services company Capgemini, which also found that many companies had abandoned efforts to to use the technology.

Paris-headquartered Capgemini conducted a primary survey of 500 automotive executives from large automotive organizations in eight countries. It then followed the survey up with in-depth interviews, it said in a report published this week.

Its findings were striking: the survey identified a sharp increase in companies not using AI at all (from 26 percent to 39 percent). According to the report, just 26 percent of companies are now piloting AI projects (down from 41 percent in 2017).

This is maybe due to “companies finding it harder to realise a desired return on investment” Capgemini said, adding that to deliver the benefits of Artificial Intelligence at scale, companies must “invest, upskill and create infrastructure”

automotive AIAutomotive AI: A Missed Opportunity

“The modest progress in implementing AI projects at scale represents a major missed opportunity for the industry” Capgemini said.

“Modelling in the report, based on one typical Top 50 Original Equipment Manufacturer (OEM), estimates that delivering AI at scale could achieve increases in operating profit ranging from 5 percent (or $232 million) based on conservative estimates, to 16 percent (or $764 million) in an optimistic scenario.

The company put tyre manufacturer Continental forward as an example; the company is using machine learning-powered visual inspections of its products.

Content from our partners
Green for go: Transforming trade in the UK
Manufacturers are switching to personalised customer experience amid fierce competition
How many ends in end-to-end service orchestration?

“We have sensibly reduced the ratio of false positives with respect to the previous systems,” said Demetrio Aiello, head of the AI & Robotics Labs at Continental. “I am very confident that if we can deploy AI to its fullest potential it would have an impact on performance equivalent to almost doubling our capacity today.”

Markus Winkler, Executive Vice President, Global Head of Automotive at Capgemini concludes, “These findings show that the progress of AI in the automotive industry has hit a speedbump… vehicle manufacturers need to start seeing AI not as a standalone opportunity, but as a strategic capability required to shape the future which they must organize investment, talent and governance around.”

“For AI to succeed, organizations will need to invest in the right skills, achieve the requisite quality of data, and have a management structure that provides both direction and executive support.”

Machine learning is, despite being a term bandied about widely, hard to execute: it is challenging to move models to production, due to a diversity of deployment environments; it’s hard to track which parameters, code, and data went into each experiment to produce a model.

As a result, most major technology companies has been building internal machine learning platforms to manage the ML lifecycle. Facebook, Google and Uber, for example, have built FBLearner FlowTFX, and Michelangelo respectively to manage data preparation, model training and deployment in contained environments.

Even these, as San Francisco-based Databricks’ Matei Zaharia puts it, are limited: “Typical ML platforms only support a small set of built-in algorithms, or a single ML library, and they are tied to each company’s infrastructure. Users cannot easily leverage new ML libraries, or share their work with a wider community.”

(His company has put out an open source, cloud agnostic toolkit called MLFlow, that allows organisations to package their code for reproducible runs and execute parallel experiments, across any hardware or software platform. It integrates closely with Apache Spark, SciKit-Learn, TensorFlow and other open source ML frameworks.)

Websites in our network
Select and enter your corporate email address Tech Monitor's research, insight and analysis examines the frontiers of digital transformation to help tech leaders navigate the future. Our Changelog newsletter delivers our best work to your inbox every week.
  • CIO
  • CTO
  • CISO
  • CSO
  • CFO
  • CDO
  • CEO
  • Architect Founder
  • MD
  • Director
  • Manager
  • Other
Visit our privacy policy for more information about our services, how New Statesman Media Group may use, process and share your personal data, including information on your rights in respect of your personal data and how you can unsubscribe from future marketing communications. Our services are intended for corporate subscribers and you warrant that the email address submitted is your corporate email address.
THANK YOU