by Bernard Marr
The high level of road casualties we see each year shows that human cognitive and motor skills are not ideally suited to the task of piloting one-ton hunks of metal, at speeds exceeding 100 kilometers per hour, in close proximity to hundreds of others attempting to do the same thing. In theory, machines can react far more quickly and safely, and communicate between themselves far more effectively. This has the potential to save many lives.
Giving cars the ability to “learn” how to safely navigate is dependent on gathering large volumes of data. This can be done under simulated conditions but information gathered from the real world is likely to contribute to a better understanding of reality and therefore be more valuable, though expensive and possibly dangerous to collect.
There still exists a healthy scepticism in public opinion over the safety of autonomous vehicles. Until there is enough data to effectively counter this, politicians and legislators are likely to be extremely cautious when it comes to creating a legislative framework for their operation.
Sources
CNBC, Traffic deaths edge lower, but 2017 stats paint worrisome picture: https://www.cnbc.com/2018/02/14/traffic-deaths-edge-lower- but-2017-stats-paint-worrisome-picture.html
Wired, Tesla's Favorite Autopilot Safety Stat Just Doesn't Hold Up: https://www.wired.com/story/tesla-autopilot-safety-statistics/
50
Volvo: Using Machine Learning To Build The World's Safest Cars
Volvo Cars, based in Sweden, has a reputation for producing vehicles with a great record for safety. Recently, it announced that, from 2019, all new models will be either fully electric or hybrid – becoming the first major manufacturer to set a date for the total phase-out of internal combustion engines.
In 2010, Volvo Cars was acquired by Chinese conglomerate Geely Holding Group from the Ford Motor Company, which bought it from parent company AB Volvo in 1999. Like every other major car maker, it is betting heavily on autonomy – Volvo has said it plans to have level 4 autonomous cars on sale by 2021.
What Problems Is Artificial Intelligence Helping To Solve?
Volvo knows that understanding how its cars are used and the conditions they are driven in is key to keeping its reputation as a world leader in terms of safety. And safety is also a key driver in the move towards autonomous, self-driving vehicles.
Additionally, roads in the developed and developing worlds are becoming increasingly congested, and carbon emissions from petrol and diesel vehicles are major contributors to pollution and man-made climate change.
How Is Artificial Intelligence Used In Practice?
From 2015, Volvo conducted predictive, machine learning-driven analytics across petabyte-scale datasets gathered from its connected vehicles. It developed its Early Warning System, which analyzes over 1 million events every week to work out how they are relevant to breakdowns, accidents and failure rates in its vehicles.
In one pilot project, which ran until 2017, 1,000 cars were fitted with sensors to detect driving incidents and monitor conditions. The goal was to gain a better understanding of how vehicles and drivers react when faced with hazardous conditions such as icy roads.
Another focus of the data and analytics strategy is passenger convenience. This means monitoring uses of applications and comfort features to understand what functions drivers are finding useful and what is being underused or ignored. Volvo director of business intelligence, Jan Wassen, told me: “We are looking into what types of applications are being used and we continuously measure this in order for us to understand what it is that customers want us to develop in the future.”
What Technology, Tools And Data Were Used?
Data gathered from sensors attached to Volvo cars is uploaded to the Volvo Cloud and also shared with the Swedish highway authorities. Data analytics is carried out in partnership with Teradata.
To develop systems needed for self-driving cars, Volvo has also partnered with Nvidia and Autoliv, the world's largest car safety supplier. Together with Autoliv, Volvo established Zenuity, a software development group focusing on building autonomous driving systems with an emphasis on safety.
The systems will use deep learning to learn how to recognize and react to objects around the car based on data from cameras and sensors attached to the car. It is programmed to take all of the data gathered by these sensors and construct, in real time, a “situation map”, giving the artificial intelligence processors a 360-degree view of the car's environments. It also incorporates GPS and high-resolution map data to plot the most efficient route between destinations. It has been described as a complete software stack for automated driver-assisted systems and automated driving, with inbuilt algorithms for computer vision, sensor fusion, decision making and vehicle control, as well as connectivity with cloud applications. The real-time systems require enormous amounts of compute power, and rather than building their own processing systems in-house, Zenuity relies on high-performance-computer-as-a-service through a partnership with Dell EMC and VMWare.
What Were The Results?
Volvo's commitment to predictive analytics has enabled it to more quickly and accurately understand the faults and errors that can occur within its connected cars. This means that remedies – such as servicing or repair centers – can better anticipate the tasks they will face and their need for replacement stock.
And its ventures into autonomous driving are also bearing fruit with the launch of its Drive Me trial, which will soon see Volvo XC90 SUVs capable of self-driving delivered to trial customers in Gothenburg, Sweden, as well as locations to be announced in China and the United Kingdom. The users will drive the cars manually in everyday conditions, but will also be able to switch them to full autonomous mode on 31 selected miles of the city's roads.
Key Challenges, Learning Points And Takeaways
Volvo Cars, in line with every other major auto manufacturer, firmly believes that the future of cars lies in autonomy, and deep learning is the key to making it a reality.
Rather than huge fleets of self-driving vehicles suddenly appearing on our roads, the rollout is likely to be gradual – with vehicles becoming gradually more autonomous before fully self-driving cars become the norm.
Safety is considered to be one of the key benefits of autonomy in cars – if properly trained they should drastically reduce the number of accidents caused by human error.
Sources
CNBC, Geely's Volvo to go all electric with new models from 2019: https://www.cnbc.com/2017/07/05/geelys-volvo-to-go-all-electric-with-new-models-from-2019.html
Cio, The rubber hits the road with AI implementations: https://www. cio.com/article/3297496/analytics/the-rubber-hits-the-road-with-ai- implementations.html
Motor Authority, Volvo delivers first self-driving cars to families in Drive Me project: https://www.motorauthority.com/news/1108300_volvo-delivers-first-self-driving-cars-to-families-in-drive-me-project
Volvo Cars, Volvo Cars and Autoliv team up with NVIDIA to develop advanced systems for self-driving cars: https://www.media. volvocars.com/global/en-gb/media/pressreleases/209929/volvo-cars- and-autoliv-team-up-with-nvidia-to-develop-advanced-systems-for- self-driving-cars
Part 6
Final Words and Artificial Intelligence Challenges
51
Final Words And Artificial Intelligence Challenges
We hope that the practical examples of how artificial intelligence (AI) is being used in the real world have given you a solid overview of the current state of the art, as well as inspired you to explore the applications and implications of AI for your own career, business or industry.
We believe that the race is on for everyone to grab the opportunities AI is offering. We also believe that if you don't, then you face a very real risk of being left behind in this AI gold rush. Based on our experience working with many of the world's leading companies, from AI trailblazers, to long established incumbent organizations, to start-ups, there are a few challenges that need to be addressed to make the AI ride as smo
oth as possible. Let's explore some of the key challenges before we bring this book to an end.
Approach Artificial Intelligence Strategically
Make sure you approach AI strategically and don't apply AI to an outdated business model. You will have seen many examples in this book where AI has enabled businesses to completely reimagine and transform business models or even entire industries. In our advisory work with leading companies and governments, we have seen first-hand how important it is to develop a good AI and data strategy in which you identify the most important business opportunities and challenges AI can help you address. Once you have reached agreement on the AI strategy, it is so much easier to successfully roll out AI that will deliver real business results.
Develop Artificial Intelligence Awareness And Skills
There is a massive lack of AI understanding and a big war for talent. If the people in your organization, from the boardroom to the shop floor, don't understand what AI is and what it could do for your business, then it is unlikely your company will thrive in the fourth industrial revolution. On top of that we are seeing a global shortage of AI talent where the really skilled people command rock-star salaries and are usually snapped up by the AI trailblazers.
The lack of AI talent means many companies are outsourcing their AI projects to consulting companies. Many of the leading consulting companies offer great services around AI but it is important to realize that companies must also boost their in-house skills and capacities. AI will become such an important factor of competition that simply outsourcing it can leave companies vulnerable in the long run. A better model is to focus on developing core skills in-house and then bring in external expertise to boost capacity and ensure you transfer skills and expertise back into your company.
We have been working with businesses and governments to help them boost their AI understanding, data literacy skills and data science capacity, and have experienced the difference this can make. Once people understand the art of the possible and have access to in-house skills to turn ideas into practice, AI will quickly flourish.
Secure The Right Data
Data is the raw material for AI. Better data means better AI algorithms. The companies that have the best data will quickly gain an advantage over their rivals. It is therefore important to treat data as a vital business asset and identify which data your business really needs. Businesses must ensure they have access to the data they need, both in terms of intellectual property rights and in terms of legal and privacy rights. Those businesses that create a data strategy to identify the critical datasets they will need and then make sure they are able to collect and use that data for their own advantage will lay the foundations for their AI success.
Update Your Technology And IT Systems
Successful AI adoption requires modern technology in terms of data storage and processing power. The reason many of the AI trailblazers and AI start-ups are able to move so quickly is because they have the modern AI technology infrastructure in place, which they often built from the ground up. Companies need to be able to collect, store and process data to make the most of AI. Siloed data storage and outdated IT infrastructure are often key barriers for incumbent organizations. The companies that will thrive in the fourth industrial revolution are those that combine modern data clouds with Internet of Things (IoT) and edge computing technology so they can make the most of AI.
Use Artificial Intelligence Ethically
AI gives us tremendous opportunities to do good, but as with all technological innovations, it can be used for good and bad, and can be used well or not so well.
Businesses must ensure they use AI well and to benefit people and society and not to exploit people or use AI against them. There are many ethical questions we still have to answer about AI. For example, when we use AI in autonomous cars, should the algorithms prioritize protecting the passengers over other road users? In an emergency, how would the vehicle decide if it was better to save the life of the passenger or the life of a child waiting at a bus stop? What are the ethics around AI in weapons? We could argue that having autonomous robots, tanks and drones to fight wars would mean fewer casualties among our armed forces, but should we really give AI the power to kill people without human intervention?
Businesses need to address the ethical challenges and ensure their use of AI is as transparent as possible. They also must make sure their AI is free of biases and doesn't discriminate against certain people. Using real world data to train AI can introduce the same biases that plague our human decision making. Microsoft learnt the hard way when their AI Twitter chatbot started to mimic other Twitter users by becoming racist and abusive.1
Companies need to ensure their AI is free from biases and discrimination and must put more effort into explaining AI decision making. Deep learning AI can often be black boxes that make decisions we as humans can't understand or trace back. Understanding why a Facebook algorithm served you a Walmart ad over a Target ad might be less important but if AI suggests removing your liver or sending someone to jail, then we need a little more information on how that conclusion was reached.
In 2016, several industry leaders, including Amazon, Apple, Google, IBM and Microsoft, joined together to create Partnership on AI to Benefit People and Society to develop and share best practices, advance public understanding, provide an open platform for discussion and identify aspirational effort in AI for socially beneficial purposes. Today, the partnership has over 80 organizations across 13 countries as their members.
Prepare Yourself For Artificial Intelligence Disruption
AI will have a profound impact on jobs. As with all the previous industrial revolutions, this fourth AI-led revolution will automate many tasks that are today done by people. The difference is that this time it is not just low-skilled jobs and tasks such as taxi drivers and the jobs at supermarket checkouts that are at risk. AI has the potential to take on many jobs and tasks that today are done by highly skilled professionals such as accountants, lawyers and doctors. Even if AI doesn't take away your job, it augments most jobs.
A good way to prepare for this is to break down your own job into tasks that could be automated today or in the near future, and those that AI is unlikely to be able to do. It is then a good idea to focus on the skills where there is less AI competition and where we can provide the uniquely human touch. Skills such as empathy, social communication, critical and strategic thinking, creativity, high dexterity, imagination and visioning are all areas where humans outperform AI.
The AI revolution will also create entirely new jobs and the demand for AI and data science-related skills is likely to skyrocket for the foreseeable future. Machine learning engineers, data scientists, cloud architects, machine vision experts, natural language engineers, IoT architects, data translators, blockchain developers and data security experts are all new jobs that will be in very high demand over the coming years.
AI is going to significantly disrupt the job market and challenge our current ways of working. In the end we are the architects of our future. It is in our hands to create a world we want to live in and one that is better for us as human beings. If we are really honest with ourselves, then we will quickly realize that some of the job tasks AI can alleviate us from are not really tasks human beings should be doing – just think of jobs like putting together tax returns, trawling through massive amounts of legal cases, copying and re-entering data, etc. And we somehow have an obligation to pass tasks to AI if it can do them better and more reliably than humans – think of identifying anomalies in MRI scans to detect cancer, translate documents from English into Chinese, etc.
AI has the potential to make our world a better place, but to get there we need to make the right decisions and address some big challenges and obstacles. With this we would like to pass the baton over to you and hope this book has given you enough inspiration to start your journey.
Connect To Keep The Conversation Going
Finally, please stay in touch and keep t
he conversation going. We share a lot of content across our social media channels and the www.bernardmarr.com website.
Let us know your thoughts or questions, we are always on the lookout for new case studies for our Forbes column, and please get in touch if you feel we could help your organization along the journey.
Here are some of the ways you can connect:
Website: www.bernardmarr.com
LinkedIn: Bernard Marr
Instagram: Bernard Marr
Twitter: @bernardmarr
Facebook: BernardWMarr
YouTube: BernardMarr
eMail: [email protected]
Notes
1The Guardian, Tay, Microsoft's AI chatbot, gets a crash course in racism from Twitter: https://www.theguardian.com/technology/2016/mar/24/ tay-microsofts-ai-chatbot-gets-a-crash-course-in-racism-from-twitter
About the Author
Bernard Marr is an internationally best-selling author, popular keynote speaker, futurist and strategic business and technology advisor to governments and companies. He helps organizations and their management teams prepare for a new industrial revolution that is fuelled by transformative technologies like artificial intelligence, big data, blockchains and the Internet of Things.
Bernard is a regular contributor to the World Economic Forum, writes a weekly column for Forbes, and is a major social media influencer, with LinkedIn ranking among his top five in the world and number one in the United Kingdom. His 1.5 million followers on LinkedIn and strong presence on Facebook, Twitter, YouTube and Instagram give him a platform that allows Bernard to actively engage with millions of people every day.