• mobile_menu_1_icon

    Individuals

    • Certifications

    • Programs

  • mobile_menu_2_icon

    Enterprises

  • mobile_menu_3_icon

    Resources

  • mobile_menu_4_icon

    About

Mobile Header Background
Desktop-Menu-bar

Using AI to Advance Science

12748
single-featured-image
By John White Last Updated on Jun 15, 2021

On a recent trip to a Dock facility in Dublin, they showcased an AI-based tool to mention new recipes after having crunched through thousands of present recipes to come up with potential new flavors. Though the work is interesting, perhaps such an approach is far more interesting when applied to innovation itself.

A growing percentage of innovation today is what’s known as recombinative, and sees existing ideas and concepts applied in fresh and innovative ways. Researchers are increasingly deploying AI-based tools to help devise some of these possible roads to explore.

Suggested Read: Top artificial intelligence technologies

For instance, a current paper from researchers at Carnegie Mellon University and the Hebrew University of Jerusalem highlights an AI-driven method to mine databases of patents and research papers for ideas that can be recombined into solutions for new difficulties.

Central to the approach was an attempt to find analogies that connect seemingly disparate methods and problems. They used crowdsourcing to understand how people form similarities, before using this to train a deep learning algorithm to then mine knowledgeable databases for potential innovations.

“After decades of efforts, this is the first time that anyone has gained power computationally on the analogy problem at scale,” the authors say. “Once you can search for correlations, you can crank up the speed of innovation, If you can quicken the rate of innovation, that solves a lot of other problems downstream.”

AI data science maturity model

It might suffice to say that discovering analogies is not the easiest thing for a computer to do, as it’s often one of those things we humans do without always understanding how we do it. Prior attempts at automating the job first required a handcrafted data structure to be formed, which is exceptionally time-consuming and therefore not at all accessible.

 Into the Crowd

The team turned to Mechanical Turk to recruit several helpers to look for analogous products via the product invention website Quirky. They did this by looking for a range of products with similar purposes or means of doing their job.

Also read: Most popular AI models

Each participant was told to explain how they came to their assumptions. For instance, they might describe a product in terms of both its determination and its mechanism. This helps to provide connectors that can be used in creating the analogies.

These insights were then fed into the algorithm before it was set loose on an additional dataset of product descriptions. The researchers tasked the algorithm with coming up with certain analogies of its own. Interestingly, it proved rather good at the task and went beyond mere surface similarities to discover similarities between seemingly different products. Indeed, when these were tested, the new product suggestions were rated as the most innovative ideas.

The team believes that the method could easily be scaled up to be used as part of the innovation process by letting organizations to find earlier hidden connections between patents or research papers. Both datasets are enormous and growing at a tremendous rate. This is a problem well-suited for autonomous methods, so it will be interesting to see what comes next for the team and their system.

A similar project was documented in a recently published paper in Nature. The researchers used a similar approach to derive connections between words and subjects, before trawling through around 3.3 million scientific abstracts from papers published between 1922 and 2018 on materials science.

Featured article: Top 5 jobs in AI

The method was able to capture fundamental knowledge within the field, such as the nature and assembly of the periodic table and the way chemicals’ structures transmit to their properties, whilst also suggesting a few new chemical compounds with properties similar to those of thermoelectric materials that hadn’t been examined before. The team believes this method could be an effective and efficient way of discovering new avenues for scientific research and therefore accelerate the advance of science in numerous fields.