Alibaba and Microsoft AIs Outperform Humans in Reading Test

Machine learning is one of the more interesting industries to keep an eye on right now. Various technology giants have shown a great interest in this technology, with varying degrees of success. Both Alibaba and Microsoft seem to be capable of developing “smarter” machine learning models. So smart, even, that their creations have surpassed human scores on a reading test. It’s a rather surprising turn of events, although there is still plenty of work to be done.

Taking Machine Learning to the Next Level

Most technology enthusiasts are concerned about how smart machine learning products will become over time. Although these products are coded by humans, there is no doubt these new creations will eventually outsmart our species and perhaps even take over the world one day. For now, such apocalyptic scenarios haven’t come true, but that doesn’t mean they are off the table entirely. That’s especially the case if the machine learning models developed by Alibaba and Microsoft are an indicator. Their creations perform far better on reading tests than humans.

More specifically, Alibaba’s Institute of Data Science and Technologies reports that its AI received a score of 82.44 when it came to answering questions without multiple choice options. For the average human test subject, the score was 82.304. Although this is a minor difference, it does show that the average machine learning-based tool is already better than humans in this regard. Microsoft scores relatively high as well, with a score of 82.65. It is evident that the human element in the equation has been officially surpassed by a computer, which is not necessarily a positive thing.

The big question on most people’s minds is what all this means for the future. While these results are promising for Alibaba and Microsoft, they are also a cause of concern for most human entities on the planet. Creating a natural-language processing AI which scores higher than actual people is an intriguing milestone, especially when considering that this field of research is still in its early stages. A lot of progress can still be made, but the foundation for a bright future has been laid already.

It is evident that interest in the world of AI development has never been higher. While these scores will not necessarily push a lot of companies into this field of research, no one can deny the numbers either. There is a genuine “weapons race” going on in the world of artificial intelligence development, especially in Asia. Enriching social media feeds and aiding in autonomous driving are two of the more popular concepts waiting to be explored at present.

As the name suggests, a natural language processing AI mimics the human comprehension of words and sentences. Based on that information, it can understand questions and formulate thorough answers. Not all of the answers will be correct, but a lot of great progress has been made in this regard already. It goes to show that these AI-driven solutions can process a huge amount of data without too many problems.

Do keep in mind these machine learning-driven solutions are still somewhat limited in the answers they can produce. Answering an objective question is not all that complicated, but other types of questions still yield a mix of unsuitable responses more often than not. It will be interesting to see how this technology is commercialized in the years to come. There is great interest in how AI can improve customer service, healthcare, and so forth. Whether or not reducing human input in these industries is a good thing will always remain a controversial topic.

You may be interested

bitcoincasino
shares26 views

50+ Bitcoins Total Jackpot at Bitcoin Casino US

Brian Evans - Feb 25, 2018

Striking it rich online is more than possible at an establishment like BitcoinCasino.us. This bitcoin casino has many games that feature a progressive pot—meaning that they get…

Alt coins
shares29 views

Decentralized Application Platforms Sit At The Heart of The Blockchain Revolution: Part 3

Brian Evans - Feb 25, 2018

In parts one and two of our series on decentralized application blockchain platforms (dApps) we examined the dominant players in the space as well as those that,…

Blockchain
shares20 views

Three Ways Your ICO Can Overcome FUD in the Market: Expert Take

Brian Evans - Feb 25, 2018

The ICO-market has a PR problem, FUD is a symptom of misunderstanding. #EXPERT_TAKE

Most from this category

%d bloggers like this: