The field of artificial intelligence (AI) has made incredible scientific advances in recent years, from enormous improvements in processing power and computational efficiency, to new insights into target identification, language, and deep learning. IBM – a leader since the beginnings of artificial intelligence research in the mid-20th century – has helped broadly inform those successes. And 2019 was a watershed year for IBM artificial intelligence research.
During the year, IBM researchers reached a new record of 175 regularly accepted papers at the top eight AI conferences, hosted the second annual Artificial Intelligence Research Week in September, and launched the AI Hardware Center to further explore next-generation artificial intelligence hardware.
The MIT-IBM Watson AI Lab, now in its second year of operation, has flourished – welcoming Boston Scientific, Nexplore, Refinitiv and Samsung as the first members of its new Membership Program. IBM has taken another major step towards mastering the language, and hired Harish Natarajan, a leading Debate participant in February 2019, at the IBM Think Conference as part of the Debater project. In November, IBM partnered with him and other participants in the Cambridge Union debate, the world's oldest debate society.
IBM researchers have also continued to focus on building and enabling artificial intelligence solutions that people can trust, enhancing artificial intelligence's ability to explain their recommendations to them through the Open Source Artificial Intelligence 360 Degree Guide. These are just some of the notable achievements of IBM artificial intelligence researchers this year.
In 2020, three themes will shape the progress of artificial intelligence: automation, natural language processing (NLP) and trust. Broadly speaking, we will see artificial intelligence systems that through automation work easier and faster for data processing professionals, companies and consumers. Natural language processing (NLP) will play a key role in enabling artificial intelligence systems to talk, debate and solve problems using everyday language. And with each of these advancements, we will witness more transparent and responsible practices in managing artificial intelligence data, through tools ranging from explanation to detection of discrepancies.
From this perspective, IBM researchers are discovering five of their annual predictions for artificial intelligence (AI) in 2020:
AI will understand more, and will be able to do more: The more data artificial intelligence systems have, the faster they will become better. But artificial intelligence needs for data can be a problem for businesses and organizations that have less data than others. This is not to say that they cannot count on artificial intelligence support. Over the next year, more AI systems will begin to rely on “neuro-symbolic” technology that combines learning and logic. Neuro-symbolic technology is a map for advances in natural language processing (NLP), which helps computers better understand human language and conversation by engaging common sense and knowledge domains. Such breakthroughs will soon help businesses deploy advanced conversion-automated customer care and technical support tools – while far less data will be required for AI training.
· AI won't hijack your job, but it will change the way you work: AI will continue to affect jobs in the coming years. And the fear that people will lose their jobs because of machines is unjustified. AI will transform the way humans work through automation. New research from the MIT-IBM Watson AI Lab indicates that AI will increasingly help us with tasks such as scheduling, but will have less direct impact on jobs that require skills such as design expertise and industrial strategy. Expect workers in 2020 to begin to notice these effects as AI begins to enter jobs worldwide; Employers need to start adjusting work roles, while employees need to focus on expanding their skills.
· AI will lead to trust in AI: In order to believe in AI, these systems must be trusted, fair and valued. We need to ensure that the public can be trusted in the technology and that its conclusions or recommendations are not biased or manipulated. Throughout 2020, reliability-regulating components will be intertwined with the AI lifecycle to help us build, test, run, monitor, and certify AI applications works confidence, not just for performance. Just as with the rise of AutoAI – the use of AI to create AI – we will witness an increase in the use of AI in AI management. This adoption will lead to the creation of reliable AI workflows in a variety of industries, especially those strictly regulated.
· The energy appetite for AI requires greener technology: Data centers are vital to the modern world – we rely on them for everything from enterprise computing to social media and playback of your favorite movies. They also support artificial intelligence and are estimated to represent as much as 2% of the world's total energy use. Demand for cloud computing and AI will not disappear, so expect efforts to make AI technology more sustainable in 2020. This includes the creation of new materials, such as “transition metal oxides” that make devices more flexible, the design of new chips and those with analog and mixed signal processing as well as new software techniques based on approximate computing, all with the aim of increasing AI workload with at the same time reducing carbon emissions.
· AI lab assistants discover new materials: For over two centuries, the synthesis of organic molecules has been one of the key aspects of industrial chemistry research. As a result, the world has life-saving drugs and synthetic fibers. Still, researchers are still struggling to investigate hundreds of thousands of possible reactions when creating different molecules. The total amount of information means that a scientist can memorize several dozen reactions in his field, but it is impossible to be an expert in everything. Now they may not have to. IBM has developed an AI tool that can predict millions of chemical reactions – both back and forth – as well as synthesize molecules in the cloud – called RXN for Chemistry (try this service today). Expect a significant increase in the leverage of AI and automation in 2020 to drive progress in material discovery and development.