By Faisal Khan | 16 September 2019
AI has the potential to disrupt every field across the whole ecosystem and it has been making huge strides in that direction. If we are able to come around the ethical concerns regarding the development of Artificial Intelligence, it can help us to perform groundbreaking research in record time – helping us reinvent the way we invent. One such application that I am really excited about is in the healthcare industry.
— TwinyBots (@Twinybots) November 9, 2019
Drug discovery is a tedious, expensive and often a frustrating process costing tons of money and thousands of precious lives before a cure for a major disease or outbreak is discovered. It’s estimated that potential drug-like molecules outnumber the no. of atoms in the solar system. Needless to say that humans can only explore a tiny share of these with the associated constraints.
This is where AI’s Machine Learning Algorithms are so good at – traversing seemingly unlimited possibilities in a limited amount of time. Atomwise, Kebotix and Deep Genomics are drug-centered AI startups which are performing research based on the combination of neural networks, AI and Robotics.
Let’s review some of the other AI-based models that have been tested recently.
Researchers at the prominent Seattle-based Allen Institute have been at work trying to create a “digital Aristotle” since 2013. Last week, they unveiled a new AI system that scored more than 90% on an eighth-grade science test and more than 80% on a 12th-grade exam. This marks significant progress considering more than 700 computer scientists failed to build a similar system four years ago. The most sophisticated AI system at that time scored a paltry 60% on the test.
The World’s top AI labs have been rapidly improving to build better systems capable of understanding and responding to natural language. Machine Learning algorithms have also improved drastically to analyze documents, parse information and even generate a language of their own. Aristo can be considered the next-generation AI system.
Scientists still believe a lot more work still needs to be done. Aristo is only built to answer multiple-choice tests which don’t include pictures and diagrams. The next phase would be to design an AI system that incorporated additional skills like understanding logic and computer vision.
.@allen_ai's #Aristo #AI system finally passes an 8th-grade science test, marking "breakthrough" for artificial intelligence: https://t.co/LSBi52LSrL H/T @etzioni @VulcanInc pic.twitter.com/xiR5ZnaQcw
— Alan Boyle (@b0yle) September 4, 2019
This is the next step to the turning point in 2016 when a London Lab’s AI system beat the World’s best players at the ancient game of Go. Other Labs are currently working at building advanced neural networks that can understand and comprehend the complexities of natural language by analyzing the books written by humans.
For those of you who don’t know Open AI, it is a not-for-profit organization, founded in 2015, which conducts research in the field of Artificial Intelligence. They released an AI tool six months ago that generated text on a wide variety of subjects – fake news to poetry. These texts mimicked the writing style of humans with high accuracy.
The company has been rolling out the tool in stages, improving on its shortcomings with a newer version every time. The latest version is dubbed GPT-2, and is the most powerful version yet, accounting for 50% of the power of the full version. You can get hands-on experience with the latest tool by trying it here.
AI systems have gotten pretty good at creating music and paintings and even Renaissance-style selfies. This latest tool can do wonders for fiction writers. GPT-2 captures the human emotions in a human writing style perfectly as illustrated in this article.
On similar lines, Allen Institute has also developed an AI model called Grover, created to spot fake news written by machines. Trained on 120 gigabytes of real news articles, gathered from the 5,000 publications on Google News, it can differentiate a human-written story from a computer-version with 73% accuracy. The interesting part, however, is that creators of this AI not only taught it to spot fake news but to write them as well.
We found that just as a large transformer model trained on language can generate coherent text, the same exact model trained on pixel sequences can generate coherent image completions and samples. https://t.co/whREMuBvxx pic.twitter.com/6AGsfx26gM
— OpenAI (@OpenAI) June 17, 2020
All you have to do is to enter a headline and it will generate a fake news article consisting of hundreds of words, making you believe that it is from a leading news publication. Grover is so convincing that it can even fool itself – detecting its own generated fake news 92% of the time while fooling itself 8% of the time!
While there is much debate on whether AI systems would ever be able to mimic the empathy that humans have in-built by nature, Humana Pharmacy has ironically deployed an Artificial Intelligence system that gives its call center employees gentle nudges to be friendlier and more empathetic with customers.
The algorithm was developed by an AI firm Cogito. The system reminds people when they are silent for too long or start speaking too quickly. The algorithm works by analyzing the biological signaling mechanism, which minimizes any language and culture bias. A secondary algorithm and a human annotation team is also put in place to check for this bias.
According to USA Today, Cogito works with call centers of large insurance companies, including MetLife and Humana, retail banks and credit card issuers, and says it has more than 25,000 users.
This is not to conclude that Artificial Intelligence has reached a human level comprehension of emotions or replacement of humans in the current scheme of things. However, the transformation towards smarter AI systems does seem to be gathering pace.
Reprinted with permission from the author.
Faisal Khan is a prolific Canada-based tech blogger and influencer. He is the founder and editor of the Technicity publication which focuses on technical, scientific and financial knowledge sharing. Follow him on Twitter @fklivestolearn.
Artificial Intelligence & the Future – Rise of AI (Elon Musk, Bill Gates, Sundar Pichai) | Simplilearn
GPT-3 Demo: New AI Algorithm Changes How We Interact With Technology
Be sure to ‘like’ us on Facebook