Everybody and their neighbor (plus their neighbor’s dog) is buzzing about artificial intelligence today. It’s tempting to roll your eyes at the speed of this hype train. But, you have to admit: there’s a lot to it. We are coming very far very fast in terms of how AI can transform our personal and working lives. And it’s poised to become a key part of our day-to-day, as common and as transformative as electricity.
Perhaps the most interesting thing about AI is that, while it can automate and do “work” at greater efficiency, it uses machine learning to “think” and “learn” over time, strategizing, designing, recognizing patterns, and making decisions. If that sounds a lot like a human brain, it’s because deep learning, one of the most important methods of machine learning, is based on the idea of a neural network, modeling the structure and function of the human brain.
And the impact of all this is huge. Some estimate a staggering $13 trilllion in global economic activity by 2030 thanks to AI. And millions of new jobs are expected in the coming years. (At Springboard, we recently launched the first AI / machine learning course in the world with a job guarantee because we anticipate so much growth in this space.)
With things moving so quickly, we decided to pause and look back at some of the new AI developments over the past year or so. Not another rhetorical trend piece, but a real overview of some of the new areas and advancements to watch out for.
AI Developments for 2019
Voice Assistants Are Handling Nearly Half of Searches
It’s clear that AI assistants such as Siri, Alexa, Google Assistant, and Cortana are becoming far more useful. Powered by natural language processing, voice assistants are now handling an estimated 40 percent of all searches. And one in six Americans now owns an Alexa-compatible speaker.
AI assistants are constantly improving because they are always receiving more data from users. That enables them to get better at speech recognition and request handling. Not to mention, the major players are rapidly adding features and nurturing developer ecosystems around these assistants. They will continue to better anticipate our behaviors and understand our habits, proactively recommending actions and services for us.
On the other hand, one area where the buzz may be overdoing it is robot calling assistants. A recently hyped example is Google Duplex, which is an AI that makes audio calls. Experts think that robot calling assistants are still a bit early, so there is a ways to go before you can confidently have a virtual secretary make your calls and run your life on your behalf.
Ultimately, we’ll still need a combination of human and artificial intelligence for the foreseeable future. But chatbots and AI assistants are one of the main examples where progress and consumer adoption has been rapid. And the more data these assistants get, the faster they will improve. It’s a self-fulfilling cycle of development.
Facial Recognition Is Becoming Minority Report-Like
During the past year, we’ve heard more about facial recognition becoming as advanced as in the sci-fi action film Minority Report.
Real-time facial recognition will become far more common, particularly in public spaces and retail. The technology to combine web tracking with physical biometrics is enabling companies to rapidly understand who you are and then serve up relevant content (often advertising). Some interesting examples include:
- Major League Baseball will soon start allowing fans to validate tickets by scanning their face.
- Singapore’s new mega-mall will track shoppers and recommend deals to them, on the fly.
- The organizing committee behind the 2020 Olympics in Tokyo has said it will include facial recognition systems in order to improve security.
Not to mention that many of us now have facial recognition built into our smartphones, enabling us to effortlessly unlock and interact with them faster than ever before. For now, facial recognition seems like it will certainly make existing experiences better—faster, more personalized, and more secure.
But there are clear concerns: privacy and anonymity, for example, as well as more potential for fraud. Consider the fact that 3D printing is simultaneously becoming high fidelity and accessible. What if, for example, someone could scan your face and print a mask? Hackers have continued to prove that they can rapidly beat face ID, so it may take some time before it’s secure enough to deploy en masse.
More Businesses Are Moving AI From Pilots to Production
While much of the talk around the development of AI and machine learning focuses on consumer experiences (AI assistants, home speakers, self-driving cars), some of the best benefits are poised to reach businesses, particularly the enterprise.
Over the past couple of years, as AI hype has reached a fever pitch, many companies have been scrambling for an AI anything—a product, a strategy, or at least the appearance of engaging with the technology. In 2019, expect many businesses to move from showcasing and piloting AI technologies to soft launches and global deployments of AI platforms. Some examples:
- Financial services are now running real-time logs of thousands of transactions per second, parsing them through machine learning algorithms.
- Retailers are grabbing data from receipts and loyalty programs, then passing it to AI engines to determine how to sell products better.
- Manufacturers are using predictive technology to know what stresses their machinery and to predict when it is likely to break down or fail.
Data is more valuable than ever—some have even said that “data is the new oil.” Reams of data are vital to making AI engines work, enabling them to learn. So some traditional businesses, including in manufacturing and agriculture, are leveraging their businesses to provide data as a service. One example is John Deere, which is using their vast access to agricultural data to assist farmers.
Global consulting firm PwC also recently published their 2019 AI predictions report, which goes deeper into the priorities that businesses should consider.
AI Will Not Take Our Jobs
Rest assured, a robot is not going to take your job.
There are fears left and right that we will be left jobless and hopeless as supersmart AI steals all of our jobs. Others actually paint this picture as the ultimate goal, enabling humans to live a life full of leisure. The truth is altogether more benign, and somewhere in the middle.
Fact: AI is likely to create more jobs than it will destroy. Yes, AI will automate routine work, starting in lower-skilled areas with repeatable, measurable tasks. But research firm Gartner predicts that by 2020 AI will create more jobs than it will eliminate. Often in major periods of transition due to technology, temporary job loss will occur, but it will be followed by business transformation, growth, and hiring—and “AI is likely to follow this path.”
Gartner predicts that 1.8 million jobs will be lost, most notably in manufacturing industries. But they also estimate that 2.3 million jobs could be created in education, healthcare, and the public sector. These jobs will be opened up by growth in the industries. And coupling AI with human intelligence will enable a whole new set of opportunities.
Ultimately, AI automation is a reality, and some jobs will be lost. But what is being underestimated is what Gartner calls “AI augmentation,” the opportunity for AI to work alongside human professionals, enabling people to reduce repetitive tasks and be more productive. That will help reduce the time spent on tedious tasks and create more time for creative work.
U.S. vs. China: The Race for AI Dominance
Much attention is being paid to China’s focus on AI. They’ve invested heavily in their Made in China 2025 plan, which, among other areas, declares AI a critical pillar of competitiveness. There are many reports that China is outpacing the U.S. in the number of AI startups and AI-related patents created. Some experts say that China will be the next global AI superpower.
So, the race between China and the U.S. is heating up, as the U.S. makes its own investments in AI. It’s even been dubbed the new space race.
There are a few key considerations to note here. One is that overall competition and tensions between the U.S. and China have reached a fever pitch, particularly given the recent trade war and related discussions.
An effect of this tension is a rapid set of moves to “insource” or “reshore” critical technologies. For example, Huawei is now aiming to develop its own AI processing chips in order to reduce its reliance on U.S. manufacturers like Intel and Nvidia.
The optics of working with China has also affected some U.S. technology companies very publicly. For example, Google and Microsoft both faced immense public criticism recently for doing business with China. Google’s been involved in a secret project to produce a censored search engine in collaboration with the Chinese government called Dragonfly. Its employees protested dramatically after finding this out. The company had previously resolved to not censor its search engine and this has prevented them, for years, from entering the Chinese market.
U.S. companies are also facing unique challenges compared to their Chinese counterparts when it comes to working with the government, particularly on defense. Both Microsoft and Google have come under fire for using their machine learning technology to support the U.S. military, specifically around drone technology.
Ethical and Transparent AI Is in Demand—and in Development
The past year has seen many technology companies come under fire for ethical lapses, especially around data privacy and nefarious algorithms. Continued fallout from Facebook’s involvement in the 2016 U.S. election (and many other elections around the world), scrutiny of Google’s many projects, including Sidewalk Labs, and security breaches among technology companies are, not surprisingly, leading to rising mistrust among many consumers.
Worse still, the effects of that mistrust are often magnified in the context of AI and algorithms, particularly because these are relatively autonomous computers making decisions of their own accord. Algorithms drive many parts of the modern internet experience, especially on social networks, where one’s “feed” is determined by many different signals and patterns. Some people have accused feeds of promoting content that plays to their emotions, leaning heavily toward outrage and other negative emotions. But the real concern around AI in this context is that we don’t know who, what, or how decisions are being made by the products and services that we use.
It’s the black box problem: who’s in control and what are the factors that are changing an experience? Because of growing skepticism, there’s a greater onus on companies and demand for technical and transparent AI. Luckily, there is a lot of work being done to ensure that AI is more ethical going forward.
The first is around that black box issue. The goal is to make the workings of AI easier to understand, building trust, not only to reassure the public, but to improve the AI itself by exposing bias in data and algorithms. Recently, IBM improved decision traceability of its AI through OpenScale technology. The company now provides real-time insights into what decisions are made—and also how they are made, what data is used, what is weighted, and potential bias.
Another way that AI development is being improved is through data control. It’s a change that’s largely been necessitated due to government regulation, including the recently passed General Data Protection Regulation. GDPR has given citizens some protection against decisions that have a significant impact on their lives, including decisions made solely by machines. It gives citizens legal protection and it made major international companies in particular adhere to better standards.
It’s increasingly clear that we need a new social contract between people and AI, if development and adoption are going to continue positively. People need to feel that AI is fair and that they won’t be singled out or targeted. They need to trust it. To that effect, experts and human rights activists from around the world collaborated on a universal declaration that aims to protect the rights to equality and non-discrimination in the development of AI.
Another positive development is that AI might start to be accredited for ethics. According to a recent Forbes article, “the Institute of Electrical and Electronics Engineers (IEEE), which has 395,000 members in 160 countries, has launched an initiative on the ethics of autonomous and intelligent systems. One of its goals is to help consumers to identify which AI products incorporate ethical aspects and to have the information to take responsibility for their choices.” And they go on to compare the initiative to food labeling in supermarkets. Imagine being able to pick which products you use and which AIs you engage with, based on their ethical rating.
Where to Learn More
AI is an exciting and rapidly changing field. If you’re interested in learning more, check out some of these recent posts:
- 20 AI, Machine Learning, and Deep Learning Experts that You Can Follow
- 20 Machine Learning Data Science, and AI Newsletters
- 5 Untraditional Industries That Are Leveraging AI
Exploring a career in AI? Consider Springboard’s Machine Learning Engineering Career Track, the first of its kind to come with a job guarantee.