Jacek Chmiel, Global Chief of Consulting & Technology at IT Services Competence Platform:
The fragmentation of frameworks and limited interoperability in AI space is hurting its adoption. I see Open Neural Network Exchange as a key technology for broader adoption of AI in 2019. Of course, it’s going to be interesting how the big AI leaders will behave in practice, but there are positive signs for the future; time will tell how much adoption and support ONNX will receive. Another key challenge and simultaneously a trend for AI is explainable AI – AI makes decisions, but the “explaining why” part of the technology has been missing. Now solutions for this problem are on the rise. That will certainly open more doors for AI applications. I’m keeping my fingers crossed for feature design automation attempts (and AutoML in general) as a means to minimize barriers to entry for business people and the less AI-savvy part of the tech community.
Lyubomyr Senyuk, CTO and Head of R&D at CoreValue:
Now there is a tendency to loudly name anything that can produce recommendations based on God-knows-what rules, trained on tons of data, as “artificial intelligence”. But “true AI” is actually the opposite. That is, an intelligent system should use a thousand times smaller volumes of data for training, which is hypothetically possible if AI actively reuses knowledge patterns from different domains. Reusing, collecting and structuring those patterns is probably the key to building hierarchical “knowledge” systems that can be compared to a human manner of explaining things. The current AI approaches are far from that, but luckily Moore’s Law is still capable of finding ways to work, and support more and more complex approaches within deep learning (like Capsule Neural Networks). Quite some time ago, I used to say that an artificial creature able to walk as elegantly as a real horse will become the precursor of true AI’s emergence. Now that the Atlas robot is doing backflips, this milestone can be marked as completed, and in 2019-2020 we should definitely expect good self-educating robotic systems like exoskeletons, robotic prostheses and, of course, military robots.
Roland Guelle, CTO at Sevenval:
For the most part, AI is still far away from daily use cases in IT projects. In 2019 this is likely to change. With toolstack growth, usage is getting easier, and the need for research tasks will be eliminated. At wao.io (Sevenval’s automated web analytics and optimization service), we’ve improved our image optimization with machine learning in 2018. So today we’re getting much better results with compression of images (high quality and small file sizes). In addition to optimizing images for websites, we learned that AI helps include expert knowledge into technical decision-making: A sufficient amount of data allows us to improve and optimize the results.
The next blog post in the Tech Trends series will be packed with insights on the subject of data. Stay up to date by following us on LinkedIn.
Pursuant to art. 13, sec. 1 and 2 of the Regulation of the European Parliament and the Council of the European Union (Regulation (EU) 2016/679) of April 27 2016 with respect to the processing of personal data and the free flow of such data, as well as the repealing of Directive 95/46/WE; (general data protection regulation, referred to herein as GDPR) (EU Official Journal UE L119/1), we inform you that:
1. The Administrator of your personal data is “IT Kontrakt” Sp. z o.o., with its registered office in Wrocław, ul. Gwiaździsta 66, 53-413 Wrocław, e-mail: firstname.lastname@example.org
2. Your personal data will be processed in accordance with GDPR for the purpose of sending you commercial and marketing information, in particular for the purpose of contacting you by e-mail and informing you of services, products, and events involving the Administrator and its personnel pursuant to art. 6 sec. 1 item a) of GDPR
3. The coordinator of activities related to the protection of personal data undertaken at the Administrator’s business entity is Konrad Gandziarski, e-mail: Konrad.email@example.com
4. Your personal data will not be conveyed to any third country outside the EU/EEA.
5. Your personal data specified in item 2 above will be stored for the time of the Administrator’s sending commercial or marketing information or until you withdraw your consent for your personal data to be processed for this purpose.
6. In some situations, the Administrator is entitled to convey your personal data to other recipients if it is essential for the purpose of the correct settling of your Award. In such events, we will convey personal data to three groups of recipients: persons authorised by us, our staff and co-operators who must have access to personal data to fulfil their duties, processing entities that we will entrust with personal data processing, and other data recipients, including parcel delivery companies, banks, insurers, and law firms.
7. You are entitled to demand that the Administrator grant you access to personal data which concerns you, as well as correct or remove it, or limit the processing or moving thereof.
8. You are entitled to withdraw your consent at any time by sending a declaration of your withdrawal to the following e-mail address: firstname.lastname@example.org