by Bernard Marr
This was essential as it helped the theme park staff and management, who weren't used to large-scale technological disruption of their working processes, to buy into the concept.
Disney worked with outside partners, including Accenture, HP and Synapse, on the MyMagic+ program.
What Were The Results?
Visitors to the Magic Kingdom can cram a lot more into their day and go home with better memories (and more souvenirs and merchandise) if the “friction” can be removed from their trip. Less waiting in line leads to happier customers who are more likely to return.
Disney says that uptake of the services offered through MyMagic+ and MagicBands has been high, with 80% of visitors using the technology to make reservations for their rides.8
Key Challenges, Learning Points And Takeaways
Disney uses its theme parks to bring its characters and movies into its visitors’ real lives. It hopes this will make them bond more closely with its brands and franchises, and continue to buy its movies and tie-in merchandise.
This is more likely to happen if they have a good experience during their visit – advanced, intelligent analytics can vastly simplify managing the movement of huge numbers of people.
Among the many arms of the Disney corporation, the parks division was known for being risk averse with new technology.9 Overcoming this was a challenge for the team that built the MyMagic+ initiative.
Notes
1Fast Company, The Messy Business Of Reinventing Happiness: https:// www.fastcompany.com/3044283/the-messy-business-of-reinventing- happiness#chapter-Discovery_Island
2Disney Research: https://www.disneyresearch.com/
3Disney Research: https://www.disneyresearch.com/innovations/ denoising/
4CNBC, Watching you, watching it: Disney turns to AI to track filmgoers' true feelings about its films: https://www.cbc.ca/news/technology/disney- ai-real-time-tracking-fvae-1.4233063
5Wired, Disney's $1 Billion Bet on a Magic Wristband: https://www. wired.com/2015/03/disney-magicband/
6USA Today, Disney parks tech upgrades make visiting more convenient: https://eu.usatoday.com/story/travel/experience/america/theme-parks/2018/02/27/disney-parks-magicbands-fastpasses-app/374588002/
7Fast Company, The Messy Business Of Reinventing Happiness: https:// www.fastcompany.com/3044283/the-messy-business-of-reinventing- happiness
8USA Today, Disney parks tech upgrades make visiting more convenient: https://eu.usatoday.com/story/travel/experience/america/theme-parks/2018/02/27/disney-parks-magicbands-fastpasses-app/374588002/
9Fast Company, The Messy Business Of Reinventing Happiness: https:// www.fastcompany.com/3044283/the-messy-business-of-reinventing- happiness#chapter-Discovery_Island
22
Instagram: Using Artificial Intelligence To Tackle Online Bullying
Instagram has only been around since 2010 but already there's a generation that probably can't imagine life without it.
The Facebook-owned social network, which focuses on image and video sharing, has 1 billion active users as of June 2018,1 posting 95 million pictures every day.2
Recognizing that bullying, harassment and abuse are unfortunately frequent occurrences online, Instagram has announced that it is rolling out artificial intelligence (AI) to stop unpleasant behavior in its tracks before it affects people's lives.
What Problems Is Artificial Intelligence Helping To Solve?
Behind all those sun-kissed holiday selfies and stunning scenery, there's a darker aspect to social image sharing.
UK charity Ditch The Label's annual bullying survey found that 42% of young people had experience of being cyberbullied on Instagram3 – the highest figure for any social platform.
In its report Cyberbullying's Impact on Young People's Mental Health, the Children's Society found early in 2018 that “the steps being taken by social media companies in response to cyberbullying are inadequate and inconsistent”.4
Like real bullying, cyberbullying can have a deep and lasting impact on its victims’ lives, in extreme cases leading to mental health issues and suicide.
By taking responsibility for keeping their users safe from bullying and harassment, social media companies have to walk a tightrope between restricting freedom of speech and policing users’ content.
How Is Artificial Intelligence Used In Practice?
Instagram rolled out artificially intelligent comment filtering algorithms, which screen all comments uploaded to the network.5
The filter is switched on by default on all accounts, but can be deactivated if someone really feels they need an uncensored experience.
Text, as well as speech in videos, is parsed by the filter, and anything that is determined to be an abusive comment – insults about someone's appearance, or their race or gender, for example – is automatically filtered.
When particular accounts are highlighted as being frequently filtered, it will trigger a manual review from the network's human staff, who will determine if the user behind the account is breaching its terms of service. This could lead to them being barred from using the platform.
What Technology, Tools And Data Are Used?
Instagram's anti-bullying filter uses natural language processing technology called DeepText, which was developed by Facebook.
It works by examining the text that makes up users’ comments and working out if they show patterns that fit with other uploads that have been flagged for abuse.
DeepText uses deep learning involving neural networks to classify the text used in uploads, as well as the context set by the text. Because deep learning systems improve in accuracy as they are trained, it becomes increasingly able to distinguish, for example, insults thrown between friends in jest and language that is indicative of a targeted campaign of cyber-harassment.6
In common with other natural language-based deep learning systems, it is able to learn and adapt to the way humans communicate in text to become increasingly good at understanding slang, patterns of speech, regional language variations and turns-of-phrase.
Where Facebook says that it is breaking new ground with DeepText is in additional layers of meaning it assigns to each word as it carries out its analytics.
As well as assigning an identifying tag to each word, and using it to track a word's frequency and context in a piece of text, it assigns each word a position on a web of semantic connections.
These allow the AI to learn about common relationships between words, and situations where different words are used to mean the same thing.
The trick is that it can do this so fast that the system is effective in real time. Which means analyzing, understanding and making a decision about 1,000 Instagram uploads per second.
What Were The Results?
Instagram's anti-bullying initiative is very new and the company hasn't spoken about the results it has seen yet.
However, the hope is that by removing offensive, upsetting or hurtful comments before they can be seen, users will have a more positive and inclusive experience on the platform.
Key Challenges, Learning Points And Takeaways
Bullying is a problem that's always existed in society, but the internet and social media make matters worse as victims can be targeted publicly and anonymously.
Without AI it wasn't possible to screen every upload to Instagram in real time. This meant that proactive blocking as enabled by DeepText wouldn't be possible.
There's also less risk of inadvertently infringing on someone simply exercising free speech or their right to object or disagree. Text analytics and natural language processing are advanced enough now to reliably make the right call.
Notes
1Statista, Number of monthly active Instagram users from January 2013 to June 2018 (in millions): https://www.statista.com/statistics/ 253577/number-of-monthly-active-instagram-users/
2Sprout Social, 18 Instagram Stats Every Marketer Should Know for 2018: https://sproutsocial.com/insights/instagram-stats/
3Ditch The Label,
Anti-Bullying Survey 2017: https://www.ditchthelabel .org/wp-content/uploads/2017/07/The-Annual-Bullying-Survey-2017-1 .pdf
4The Children's Society, Cyberbullying's Impact on Young People's Mental Health: https://www.childrenssociety.org.uk/sites/default/files/social-media-cyberbullying-inquiry-summary-report.pdf
5Instagram, Protecting Our Community from Bullying Comments: https://instagram-press.com/blog/2018/05/01/protecting-our- community-from-bullying-comments-2/
6Facebook, Introducing DeepText: Facebook's text understanding engine: https://code.fb.com/core-data/introducing-deeptext-facebook-s-text-understanding-engine/
23
LinkedIn: Using Artificial Intelligence To Solve The Skills Crisis
LinkedIn has built a social network for the professional world. What Facebook has done for keeping us in touch with friends and family, LinkedIn replicates for our working lives.
So, while Facebook primarily makes its money selling our data to businesses so they can advertise products to us, LinkedIn's revenue comes from employers looking to entice us to join their ranks.
Instead of categorizing us by what movies or music we click “like” on, it looks at our job skills and experience. Then it uses artificial intelligence (AI), which is baked into every feature on its platform, to match us with opportunities or bring us to the attention of prospective employers.
What Problem Is Artificial Intelligence Helping To Solve?
Matching applicants to job roles is a challenging and expensive task for businesses – according to Glassdoor, the average hire in the United States costs a company $4,000.1
That's a fair chunk of money even if you get lucky and find someone good – however, evidence seems to suggest humans aren't particularly great at selecting the right person for the job.
A study last year by the UK Recruitment and Employment Confederation found that businesses were failing to make the right hires for two out of every five roles.2 It also calculates that an unsuccessful hire at middle management level can cost a company an average of £132,000.
Part of this inefficiency is that recruiters have traditionally been able to gather little information during the recruitment process. Often all they have to go on will be a person's CV, interview performance and references.
In addition, particular professions are challenging to recruit for – whether it's due to lack of skills or lack of willing applicants within a geographic area. Teaching is one example where 100,000 US classrooms began the 2016/17 academic year with a teacher not qualified to teach.3
Meanwhile, according to the American Association of Colleges of Nursing, over 1 million more nurses will need to be recruited by 2024 to cope with the ageing population.4
Technology in particular faces a skills crisis. The AI revolution itself, which is sometimes predicted to cause widespread human unemployment, is actually having the opposite effect at the moment.
The rush by industry to embrace AI, coupled with the difficulty in finding candidates, means there will be 2.7 million unfilled vacancies in data science by 2020, according to IBM.5
Inefficiencies in finding candidates to fill any of these vacancies could have ruinous consequences beyond each respective industry, with knock-on effects impacting the entire economy.
How Is Artificial Intelligence Used In Practice?
LinkedIn gathers data on millions of professionals and then uses AI search tools to match applicants with jobs, and vice versa.
It also lets us build our network by suggesting people we know and might like to connect with, and uses AI analytics to identify these connections.
It even uses AI to suggest courses you could benefit from, from its library of LinkedIn learning courses.
Those looking for work have the option to mark themselves as “open candidates”, meaning they have indicated they are open to new opportunities.
LinkedIn assesses information the user has given as well as details about how they use their profile – what jobs they browse, for example – to build up a profile.6
When an employer indicates that they are looking for candidates, LinkedIn's algorithms match them with candidates who fit the profile of others who have successfully filled similar vacancies.
As the system uses machine learning, it is continuously refining its algorithms based on feedback from previous matches.
This means it can become increasingly good at predicting who will be the best candidate to fill a role. Its criteria for making the selection may be something that would go entirely unnoticed by a human recruiter with a pile of CVs to sift through.
But machine learning will build patterns in the links between the types of candidates who successfully fill different vacancies, with increasing confidence in its predictions.
LinkedIn also says that AI is a part of the process it uses to design its services. This means data-driven insights are used to determine what features and functionality users will get from the platform in the future.
What Technology, Tools And Data Were Used?
LinkedIn uses data that users give the social media account about their professional lives, such as their work experience, skills, achievements and where they are willing to relocate to.
It uses this to build a picture of which vacancies might tempt them to reply, and categorize them according to the probability that they will respond to enquires or fit a particular role.
It also builds up profiles as they use the service by monitoring what companies and vacancies they browse, as well as who joins their personal network.
LinkedIn's Recruiter platform, which businesses use to find candidates for their vacancies, also collects data on its users’ search and browsing activities.
This data is used to profile recruiters and tries to build models that predict what they're looking for in recruits.7
What Were The Results?
LinkedIn say that improvements to the AI algorithms used in its tools and search engines increased the response rate to its users’ InMail messages by 45%, and the number of conversations started between recruiters and candidates doubled, in one year.8
The social network has also announced that all of its software engineers will be trained to use AI, particularly deep learning, due to confidence that it can bring improvements to any aspect of its business where it is deployed.9
Key Challenges, Learning Points And Takeaways
Like any social network, LinkedIn's fuel is the data its users feed it, and what it can learn from their behavior as they use its services.
Machine learning algorithms can be used to accurately match candidates with job vacancies, but they need the data to be able to do it.
AI job matching can encourage employers to consider candidates with experience and skillsets that differ from their preconceptions about what a job requires.
Notes
1Glassdoor, How To Calculate Cost-Per-Hire: https://www.glassdoor.com/ employers/blog/calculate-cost-per-hire/
2The Recruitment and Employment Confederation, Hiring mistakes are costing UK businesses billions each year – REC: https://www.rec.uk.com/ news-and-policy/press-releases/hiring-mistakes-are-costing-uk- businesses-billions-each-year-rec
3The Week, America's Teaching Shortage: http://theweek.com/articles/ 797112/americas-teacher-shortage
4American Association of Colleges of Nursing, Nursing Shortage Fact Sheet: https://www.aacnnursing.org/News-Information/Fact-Sheets/ Nursing-Shortage
5IBM, The Quant Crunch: https://www-01.ibm.com/common/ssi/cgi-bin/ ssialias?htmlfid=IML14576USEN&
6LinkedIn, How LinkedIn Uses Automation and AI to Power Recruiting Tools: https://business.linkedin.com/talent-solutions/blog/product- updates/2017/how-linkedin-uses-automation-and-ai-to-power- recruiting-tools
7LinkedIn, How LinkedIn Uses Automation and AI to Power Recruiting Tools: https://business.linkedin.com/talent-solutions/blog/product- updates/2017/how-linkedin-uses-automation-and-ai-to-power- recruiting-tools
8LinkedIn, How LinkedIn Uses Automation and AI t
o Power Recruiting Tools: https://business.linkedin.com/talent-solutions/blog/product- updates/2017/how-linkedin-uses-automation-and-ai-to-power- recruiting-tools
9VentureBeat, LinkedIn plans to teach all its engineers the basics of using AI: https://venturebeat.com/2017/10/24/linkedin-plans-to-teach-all-its-engineers-the-basics-of-using-ai/
24
Netflix: Using Artificial Intelligence To Give Us A Better TV Experience
Netflix evolved from a DVD-by-mail rental company to become a subscription-based streaming video-on-demand service with 130 million subscribers worldwide.1
Netflix doesn't (yet) show advertising on its platform, but generates revenue from the subscription fees its customers pay. The driving force behind its sustainability is that customers feel they are getting good value for money for their monthly fee.
To ensure this, Netflix's TV and movie output is geared towards the concept of “binge watching” – essentially keeping customers glued to their TV sets for extended periods of time. The theory (which seems to be working out) being that this will make them feel their subscription fee is a worthwhile investment.
What Problem Is Artificial Intelligence Helping To Solve?
Consumers aren't exactly short of entertainment options these days. Between streaming movie services, the internet, video games and traditional broadcast TV, there are thousands of channels and services vying for our attention as we sit glued to our sofas during down-time.