Then I moved to a wider audience and was an evangelist for large-scale startup developers until about two years ago when I switched to the role of software engineer working with clients on some digital innovation ventures. So the theory is that a lot of companies believe that a lot is happening right now and that the world is shifting. So survival in the next few years is crucial so they’re looking for someone to guide them on what they need to do. And we at Microsoft because we have the requisite technical understanding we are demonstrating how it should be implemented to enhance business processes while solving some real-world issues. And in implementing those projects we focus on machine learning and artificial intelligence. TechNadu: What are some of your favorite Ai-backed tools to use since one of your priority is Ai? Dmitry Soshnikov: Well we are using many open source tools. There are no common instruments, so we use a lot of TensorFlow and other software and frameworks from various open source. When it comes to big scale like Ai, a lot of computational power is typically required. So when it comes to dealing with big models we are using some Microsoft-specific tools like Azure Machine Learning which helps us to train models on a scale. TechNadu: So when it comes to Ai growth, are there any stuff you’ve been trying to do with Ai that you haven’t managed to actually implement? Dmitry Soshnikov: There are a lot of difficult issues that we try to deal with and somehow we succeed. When you’re working with an Ai project, it’s hard to say if you’ve completed it or not, because it relies a lot not on algorithms but on data from the available data. And there are a lot of cases when there isn’t enough data and if you have more data you’d probably solve it, but it’s difficult with a little amount of data. One of the things about this is a project with a customer who wanted to remember different events in the football game and that was quite difficult because you can never get enough data. For instance, if you were to remember goals in the soccer game you’d need a lot of data with goals and goals that didn’t happen too often during the game it’s almost five times like that. So it’s hard to collect like thousands of positive examples, but we’ve succeeded to some degree. How that will generalize to other future games is hard to say. TechNadu: Many voices in the tech industry are calling for the introduction of some kind of rules and regulations with respect to A.I. Creation by peculiarity fears. We have heard Musk talk about that again often. So do you think these fears I created are a level A.I can ever reach? Dmitry Soshnikov: This is a truly difficult philosophical question. I don’t even think my opinion matters because each person will try to understand his or her own views in this regard as there is no indication that A.I. Will gain recognition, and no other signs we don’t know. That’s why I think we’re living in very interesting times when we could see and understand a lot about our self-consciousness, because if we can develop the self-conscious Ai it would probably mean that we are very similar “units” ourselves as well. I still think the things we are doing now are far from being conscious. I mean, they’re mathematically complicated models, really, and not more. I think this journey to this dangerous and self-conscious Ai can take quite some time and we certainly need to be vigilant when coming closer to that self-conscious Ai. So I think it’s great that there’s plenty of debate about legislation around it, because that’s certainly a very important issue. Personally I am not interested in that very much because I am a technical person. We take this subject very seriously within Microsoft. Each Ai project we do undergoes a specific A.I. Ethics checks so we don’t get interested with ventures that are not legal. And there’s a lot of discussion about “what’s acceptable,” and if, for example, the model that tries to predict something based on the picture what it should or shouldn’t take into consideration; things like that. TechNadu: How near or far we are from creating Ais which can code or support code. There are plenty of talks about developing A.I in the cybersec market. That may help to better detect malware and other attacks, but are there also concerns that some day Ais may develop malware that will be hard to detect and combat? I think that fear is very close to the general fear of some people becoming eIndividuals from Japan report receiving emails warning them of risks of coronavirus infection in their area scaring them into opening the corresponding attachments that provide payloads for malware. The actors are looking to exploit the 2019-nCovus and Irantax seasonChristmas holiday fearbreakout. They’re always swift to build a narrow-themed advertisement with all the bells and whistles that the trickery requires. # Emotet Cryptolaemus (@Cryptolaemus1) 29 January 2020 In the case of coronavirus, the scammers have produced documents using letterheads from the Japanese disability support service provider and from various public health centres. The message claims the document contains information on how to protect against infections with coronavirus, but the recipients will only get a file riddled with malicious macros that will cause a PowerShell command to get the Emotet payload and install it. The different strains of malware used in this campaign include the Trickbot data stealer which can also act as a tool for downloading ransomware. Several users report that the emails used in this campaign belong to compromised accounts, and that the perpetrators have made some effort to persuade the emails. For eg, they use Japanese in both the subject matter and document filenames to increase the recipient’s chances of becoming fooled. Even the content of the email itself has been carefully crafted to appear authentic and the email footer uses the real signature information from the spoofed health institute.