Technology 3 min read

AI Technology Now Scores 'As' on High School Science Test

In nothing short of a breakthrough for AI technology, a neural system aces a standardized 8th-grade science test using language and logic skills.

Image courtesy of Shutterstock

Image courtesy of Shutterstock

In 2015, the Allen Institute for Artificial Intelligence (AI2) partnered with Kaggle, an online community of data scientists, to organize a contest challenging anyone to design an AI smarter than an 8th grader.

One hundred seventy teams stepped up and took part in the Allen AI Science Challenge. Each group presented an AI model that tried to answer multiple-choice questions on a science test at the eighth-grade level. But, no team was able to collect the prize money.

All of the AI examinees flunked, with the most sophisticated model scoring 59.3 percent, a proof of how difficult it is for AI models to match the intellectual skills expected from 8th-grade students.

Now, four years later, the team from Paul Allen’s hometown, Seattle, is back with Aristo.”

Aristo Passes Science Test With Flying Colors

In recent years, AI has developed its abilities substantially, showing mastery over many games like chess, GO, and poker. Even defeating professional gamers and world champions at these games.

In these games, AI models draw from an existing game tree to pick the best move as a response to its opponent’s move.

In the case of scientific tests, however, AI has to demonstrate a different skill set than mere predicting ability. Picking the right answer, even when presented with multiple-choice questions, is a lot harder.

To solve science questions, AI systems need some reasoning or an understanding of how natural language works, and a bit of common sense.

An AI program called Aristo, like a Digital Aristotle, developed by the Allen Institute, shows such a level of intelligence.

Aristo is an “exciting, multidisciplinary project that aims to develop systems that have a deeper understanding of the world and are capable of demonstrating that understanding through question answering and explanation. A key metric we use is Aristo’s performance on science exams, as these questions test many of the key skills required for machine intelligence.”

So, to see if Aristo is smart enough to pass an 8th-grade science test, the AI2 team put it to the test, and the results were nothing less than excellent.

Aristo correctly answered more than 90 percent of non-diagram, multiple-choice (NDMC) questions on the Grade 8 New York Regents Science Exam. The system even scored over 83% on a science test designed for high school seniors.

“The results, on unseen test questions, are robust across different test years and different variations of this kind of test. They demonstrate that modern NLP [Natural Language Processing] methods can result in mastery on this task,” say authors of the paper.

AI2 researchers admit that their system doesn’t provide a full solution to general question-answering tasks because the questions are multiple-choice and restricted to 8th-grade level science. But the Aristo program has yielded results that represents a “significant milestone” for AI technology.

Read More: DeepMind’s AI Mathletes Flunk High School Math Exam

First AI Web Content Optimization Platform Just for Writers

Found this article interesting?

Let Zayan Guedim know how much you appreciate this article by clicking the heart icon and by sharing this article on social media.


Profile Image

Zayan Guedim

Trilingual poet, investigative journalist, and novelist. Zed loves tackling the big existential questions and all-things quantum.

Comments (0)
Most Recent most recent
You
111
share Scroll to top

Link Copied Successfully

Sign in

Sign in to access your personalized homepage, follow authors and topics you love, and clap for stories that matter to you.

Sign in with Google Sign in with Facebook

By using our site you agree to our privacy policy.