{"id":2358,"date":"2019-03-29T12:54:26","date_gmt":"2019-03-29T12:54:26","guid":{"rendered":"https:\/\/www.editage.com\/insights\/how-is-artificial-intelligence-propelling-science-and-research\/"},"modified":"2025-02-19T04:59:13","modified_gmt":"2025-02-19T04:59:13","slug":"how-is-artificial-intelligence-propelling-science-and-research","status":"publish","type":"post","link":"https:\/\/www.editage.com\/insights\/how-is-artificial-intelligence-propelling-science-and-research","title":{"rendered":"How is artificial intelligence propelling science and research?"},"content":{"rendered":"

Artificial intelligence has made in-roads into our daily lives. In fact, almost all of us have been touched by it \u2013 be it the product recommendations that pop up on our social networking platforms, or the chatbots we interact with. In simplest terms, AI refers to any machine that behaves in a way that would be considered smart or intelligent if the same behavior were to be displayed by humans. <\/span><\/span><\/span><\/p>\n

A wide spectrum of industries is looking to explore AI as it holds the promise of drastically reducing the time taken for laborious and time consuming tasks but also provide results that can mimic human intellect. AI, therefore, is of huge interest to researchers as it has the potential to change the course of scientific discovery. And while AI-powered applications in research are in a nascent stage, the realm of options and possibilities AI can open for researchers seems endless. Let us take a look at some of them. <\/span><\/span><\/span><\/p>\n

Data analysis<\/b><\/span><\/span><\/span><\/p>\n

Over the years, the ability of researchers across different fields to generate and store data has increased to an extent that we are experiencing data deluge. This, at times, makes it difficult for researchers to analyze vast amounts of data for patterns and insights. Given the deep learning techniques that AI makes available, it can play a crucial role in easing researchers\u2019 work around data analysis. According to Gadi Singer, Vice President of Intel\u2019s Artificial Intelligence Products Group, \u201cAfter you train the deep learning system with enough examples, it shows excellent ability to predict the result that will be given by the exact model. This translates to an efficiency that could turn hours or days into second.<\/a>\u201d <\/span><\/span><\/span><\/p>\n

New discoveries<\/b><\/span><\/span><\/span><\/p>\n

A combination of human intellect and AI\u2019s deep learning technique, which refers to the learning approach of AI that emulates human intellect, can help us tap previously explored areas. New machine learning methods \u201care tackling an almost-endless realm of options\u2014like all the possible mutations in human DNA<\/a>,\u201d says Singer. Thus, the automation of science could make it possible to run large experiments competently. Interestingly, AI is also being used by pharmaceutical companies \u201cto extract information from academic papers and other written materials, which can surface new hypotheses to test<\/a>.\u201d This can lead to new discovery of newer and more potent drugs, treatment methods, etc. AI, thus, manifests the potential to help researchers reach out into areas that have been perceived as challenging. \u00a0<\/span><\/span><\/span><\/p>\n

Journal processes<\/b><\/span><\/span><\/span><\/p>\n

Academic publishing is an indivisible part of conducting research. One of the concerns most journal editors and peer reviewers have is ensuring that they are able to spot incongruences in data, statistics, references, and images within the manuscripts they process. Moreover, detecting plagiarism is another important process in a manuscript\u2019s evaluation. AI can come into picture here and can assist in flagging inconsistencies with efficiency and precision.<\/span><\/span><\/span><\/span><\/p>\n

In some areas of scientific discovery and publishing, AI has made a transition from ideas to application. Let us take a closer look at them. <\/span><\/span><\/span><\/span><\/p>\n

Conducting literature reviews<\/b><\/span><\/span><\/span><\/p>\n

Data is the basis of any research, and an important step of the research process is that of reviewing previously published literature about a topic. As the volume of published literature is growing exponentially, conducting literature review usually requires a considerable amount of time. AI-based literature exploration tools are assisting researchers to perform this task quickly and efficiently. For instance, when computer scientist Christian Berger began conducting research on self-driving vehicle algorithms<\/a>, he realized he would have to sift through over 10,000 research papers as part of his literature review. This would have taken him months, but with Iris.ai<\/a>, an AI-powered tool, he accomplished this in a short period of time, as it mapped thousands of matching documents and categorized them by topic. <\/span><\/span><\/span><\/p>\n

Aiding journal processes<\/b><\/span><\/span><\/span><\/span><\/p>\n

Journals need to process hundreds of submissions, which make journals perfect for AI application. Some of the most challenging tasks of journal editors include verifying the credibility of the submissions they receive and identifying peer reviewers for the papers that seem promising. Editors are seeking assistance of AI tools to recognize suitable peer reviewers, manage submissions, and even make decisions about a paper\u2019s acceptance. Some editors have already put AI-based tools to use for these tasks. A few months ago, peer review platform ScholarOne announced its partnership with UNSILO, an artificial intelligence software company, for providing \u201cdecision support features for editors, improving the performance of paper screening with the potential to save millions of hours in peer review time<\/a>.\u201d Thus, AI is playing an important role in simplifying the challenges editors and reviewers face, making journal publication quicker and efficient. \u00a0<\/span><\/span><\/span><\/span><\/p>\n

Are there any limitations to using AI?<\/b><\/span><\/span><\/span><\/span><\/p>\n

While AI is seen as the future of science and research by many, are there any limitations to it? Some academics envision a future of automated science, but computer scientist Christian Berger warns that, \u201cBlindly using any research engine doesn\u2019t answer every question automatically<\/a>.\u201d Therefore, while AI can support scientific discovery, it would still require human insight to scour AI-generated results or analysis. Moreover, AI-powered tools and software come at a cost, and thus, these expensive tools might be accessible to few researchers. \u00a0<\/span><\/span><\/span><\/span><\/p>\n

Automated science tends to convey the perspective that the resulting inferences would be impartial, accurate, and reliable. However, there may be caveats to the liberal use of AI in some journal publishing processes. For example, discussing the use of AI in peer review process, writer and editor Douglas Heaven feels that, \u201cmachine-learning tools trained on previously published papers could reinforce existing biases in peer review<\/a>.\u201d Thus, human intervention and decision making might continue to play a vital role even though the reliance on AI may increase.\u00a0<\/span><\/span><\/span><\/span><\/p>\n

Despite the perceived challenges, AI is deemed as the future of scientific development and advancement that can lead science to reach new highs. <\/span><\/span><\/span><\/span><\/p>\n

Related reading:<\/b><\/span><\/span><\/span><\/span><\/p>\n