Chinese Version of J.A.R.V.I.S. and Agentic AI

Ant Group (a subsidiary of Alibaba and an operator of the Alipay service) has introduced Zhixiaobao, its smart “life assistant”. This new creation of the Chinese tech giant, as South China Morning Post writes, gives hope that in the near future such smart assistants as J.A.R.V.I.S. of Tony Stark from the Marvel cinematic universe will become a common phenomenon in the Celestial Empire. Through Zhixiaobao, users can access more than 8,000 digital services and 4 million mini applications using voice and text queries. For example, through a smart life assistant, you can order food, call a taxi, book a flight, analyze monthly expenses, make an appointment with a physician, track a parcel, find a restaurant to your taste, and so on. It also has built-in AI agents such as an English tutor and a fitness trainer. The smart assistant has the Bailing language model under the hood and can be accessed through a separate application or the Alipay homepage. 

Agentic AI Is predicted to become a new technological breakthrough after GenAI

Last week, a number of global media published articles about the technology, which may soon take GenAI’s place in the media headlines and minds of AI users. We are talking about the so-called Agentic AI, an artificial intelligence capable of independently performing certain tasks without involvement and prompting of a person. The New York Times, referring to some tech executives, reports that soon the development of AI tools will lead us to a kind of “efficiency nirvana” where digital assistants will anticipate and satisfy our needs even before we have to ask. The headline of Forbes even states that Agentic AI will be the next big breakthrough transforming business and technology. The VentureBeat portal is announcing a series of articles on how this technology will affect the work of organizations from various industries. The AI industry is already mastering this relatively new term – agentic: there are startups of the same name, and concepts such as agentic economy, agentic systems, and agentic products. Read this article to learn how and in what sense it is used now, and what is the irony of such use.

MIT scientists have created a tool to increase the transparency of training data for AI

Scientists from MIT, in partnership with several other institutes, have developed a tool that helps create the most suitable training data sets for LLMs. It is called Data Provenance Explorer. The essence of its operation is to analyze data that is supposed to be used to train models and automatically prepare a short report on them. This report contains information about the creators, sources, licenses, and acceptable uses of the dataset. This helps developers to make more balanced decisions on what data to use in their projects. According to scientists, such tools will contribute to the responsible development of artificial intelligence. Now transparency and quality of data for training models is one of the main challenges of the industry. Errors and biases in the datasets, as well as incomplete information regarding licenses and permitted ways of using materials, lead to legal and ethical issues. And it can also worsen the performance of the created AI solutions. Read more about Data Provenance Explorer on MIT News and Nature Machine Intelligence.

Scientists from the University of Southern California have developed a new AI algorithm that can detect brain patterns associated with certain actions against the background of all others that are happening at the same time. For example, if you are reading a newspaper, while feeling a little hungry, and at the same time reaching for a cup of coffee, this algorithm is able to identify activity in your brain that reflects specifically the movement of your hand. This new AI algorithm is called DPAD, Dissociative Prioritized Analysis of Dynamics. According to scientists, it operates more accurately than previous methods, and is able to significantly improve brain-computer interfaces designed to restore motor ability for paralyzed people. And DPAD can also detect new brain patterns that are otherwise easy to miss. To learn more, go to the USC School of Engineering website and to Nature Neuroscience. 

And finally, some news from Russia. A recent study conducted by ROMIR and Yakov & Partners showed a very high awareness of Russians in the field of artificial intelligence. 84% of respondents answered “yes” to the question “Do you know what artificial intelligence is?”, which is significantly more than in the USA (67%), UK (64%), and Germany (61%). It also turned out that almost half (47%) of the surveyed AI users in Russia prefer domestic services, and 36% use both Russian and foreign products. As for assessing the impact of artificial intelligence on the lives of Russians, the results are not so clear. Thus, 47% of respondents found it difficult to answer how technology will affect their lives, another 10% do not see any influence from AI at all. 29% of respondents expect positive changes related to AI, and 14% expect negative ones. More figures are available on the ROMIR website.