From Science Fiction To Science Journals: How AI Is Reshaping Research

21/03/2017 16:54

AI no longer belongs to the realm of science fiction. From voice recognition to self-driving cars and medical diagnosis, 'artificial intelligence' has, in just the last few years, woven its way into our lives at work and at home.

Part of this step change is due to the evolution of AI from the domain of academic researchers and tech giants such as Google and IBM to the commercial mainstream. AI today is a fully-fledged category and a rich ecosystem of startups are rapidly applying AI technologies to almost every industry and sector - including academic research.

Before we dive into how AI could reshape research, let's consider the term "artificial intelligence". It's an increasingly broad church, used to describe multiple technologies-- including machine learning, deep learning, neural networks, computer vision and natural language processing. The emergence of major AI platforms such as Google's Tensor Flow and Microsoft's Cognitive Toolkit has enabled technologists and startups to innovate without having to develop their own proprietary systems. Combined with increasing access to low-cost computational power and publicly accessible structured data, this has dramatically lowered the barrier to entry for companies, organisations and researchers.

As Dr. Stephen Cave, Executive Director of the Levelhulme Centre for Future Intelligence, said in his opening address at the recent AI in Business and Entrepreneurship Summit in London, it is possible "to reap the benefits of AI whilst preserving humanity" if researchers and developers continue to converse openly with the public and address their concerns about how AI will affect their lives.

Last year, five of the world's tech giants - Google, Facebook, Amazon, IBM, and Microsoft - announced The Partnership on Artificial Intelligence to Benefit People and Society; a new non-profit coalition between corporates, academics and the nonprofit world. The partnership aims to conduct research and recommend best practices relating to "ethics, fairness and inclusivity; transparency, privacy, and interoperability; collaboration between people and AI systems; and the trustworthiness, reliability, and robustness of the technology."

With responsible thought leadership - aiming to deliver standards of best practice that stretch beyond Asimov's three laws of robotics - we should remain optimistic about the potential benefits of AI to reshape the way we engage with technology, the world and each other. So how then might AI, the product of decades of scientific research, impact the realm of scientific research itself?

In September of last year, Google went live with its Neural Machine Translation system, a seminal project harnessing deep learning to improve translations between languages. Within months, Google's AI had generated its own internal "inter-language" that allowed it to make connections between concepts and words that had not been formally linked. With AI now capable of developing its own language and even self-programming, it's conceivable that one day, machines will be able to generate original hypotheses, conduct scientific experiments, and even write and publish the results.

At Sparrho, we see the maturation of AI technologies as the prelude to an unforeseen step change in humankind's ability to understand and innovate. We believe that leveraging human expertise to train AI can increase the pace of democratisation of science by disrupting the way current research and development is performed and communicated.

The traditional tools for scientific research can't keep up with the rate nor complexity of our modern exploration. In a world of increasingly multidisciplinary research, the conventional search approach using keywords often misses important content in tangentially related fields. Four to five thousand research articles are published daily and this number is only set to rise with new AI tools for research. Staying on top of the science that matters is increasingly difficult. To mix metaphors, it's rather like trying to drink from the firehose while looking for a needle in a haystack.

Sparrho's technology enables researchers and those with a passion for science to save time on discovering, curating and sharing scientific research important to them. Our machine learning algorithms work in the background to present the most relevant papers to each individual user from our corpus of over 48 million pieces of scientific content, without needing a lengthy list of precise keywords. In addition to aiding discovery of relevant research, Sparrho allows users create 'pinboards', curating research papers in a dynamic, more shareable way. With over 600,000 pins created and shared to date, we're proving there's demand for new, innovative ways to showcase cutting-edge science.

And we're not planning to stop there. Our vision is to evolve our technology into one that anybody with a curious mind, regardless of their academic background, can use to make their own discoveries and inventions. The Digital Age created a new class of citizen scientists: people from every walk of life with a passion for better knowledge. In the coming Age of AI, we'll give those citizen scientists unfettered access to relevant scientific research and powerful new machine learning tools. The next moonshot development will be something anyone who is passionate about science can reach for.