To contribute to the event, I held an informational talk. It detailed how to implement embedded vision into products, mentioning Luxoft’s successes and lessons learned as a cutting edge technology business. (To learn more,
I just kept hearing that word – ‘practical’.
The truth is, practical AI applications are a rarity – not many engineering teams have made AI projects that have the same level of sophistication as ours. Technology enthusiasts on the outside looking in are still curious on how to implement AI into daily life. How does and will AI affect everyday life and the actual businesses that use it? Where is AI right now on its journey to being practical? How do businesses go about adopting the technology?
After two days, this was still an ongoing topic at our booth. So after attending the event, I wanted to share 5 observations to help companies jumpstart their AI initiatives and adapt to the future.
1. Companies bridge the gap between concept and production
There’s now a bigger focus on real-world projects. For example, at the event, a large retailer was interested in using ML to optimize product placement and better serve their customers. Their project has store shelves automatically alert managers when empty, and documents how long customers spend at each shelf to determine product popularity. There was also an infant care products manufacturer present that uses AI to determine songs babies enjoy. Based on behavior, it automatically chooses which lullaby will best soothe them. There were also appliance manufacturers working on intelligent devices for consumers, with many soon in production. And a railroad company was starting to use computer vision to better understand the environment around the railroad in hopes of making it safer.
This is a stark difference from last year, when individuals were mostly focused on AI concepts – the industry of AI and computer vision has made a notable leap from concept to commercial.
2. An industry representation shift
Last year, almost all the attendees were from what I call ‘geek crowds’. They were engineers, researchers, and technical experts from startups. Of course we had them too this year, but we also had a lot of big companies from multiple industries – not just tech.
The focuses of companies are changing. When listening to some talks, I noticed many platform companies see ML as one of their top 3 priorities. Companies like Microsoft, NVIDIA and Google are making really good progress with their platforms, making them more convenient and appealing to customers. AI has matured, and businesses of all kinds want to use it to optimize company processes and get ahead in their respective industries. Whether it’s creating smart street signs or improving medical imaging, computer vision is here to stay.
3. Speaking in less math
As something that was first pioneered by scientists, AI is moving from being a scientific “buzz” area to something non-scientists can talk about and apply. The previous event had a lot of theoretical presentations, where speakers would refer to AI only in mathematical models and equations.
This year, everyone seemed to understand it both mathematically and practically, and spoke about it with less equations. The talks were from a completely different part of the value chain, and audience members understood it. The same person that’s interested in making a railroad safer via computer vision also offered frameworks on how to implement embedded vision cameras. It’s just all a result of the natural evolution of the field.
4. A desire for the practical
The talk I gave not only was understood by the audience, but they resonated with it. I briefly touched upon this earlier; when people came by the booth, they said the issue rings true – they lack practical aspects.
To make practical ML successful, it’s becoming more and more evident that data is the critical piece, and companies need to know how to obtain it. After all, ML algorithms learn by ingesting data, the only ‘food’ they consume. That makes data vital to any AI endeavor. And while replicating another’s software could take only 1-2 years, accessing someone else’s data is nearly impossible; you must be able to obtain your own customized data that fits your business’s needs.
5. Evaluation of data
There is a huge need for data scientists, as companies need them in order to thrive. But salaries for these experts have grown dramatically within the last two or three years – leaving companies wondering if they really need them at all, considering the extra cost. But really, budgeting for them is necessary. Businesses won’t get ahead without a deep understanding of their data, which is what data scientists excel at.
With a large team of engineers and data scientists, Luxoft has helped a leading agriculture manufacturer use embedded sensors to increase customer appeal, improving the efficiency of their products. We also developed a monitoring system for a fleet of sensor-equipped buses, allowing our client to utilize predictive maintenance to help prevent faults. AI is the future, and data experts help drive that future.
AI has earned its place in the real world by becoming a part of practical applications, both now and beyond. But since this was only a 2-day event, there were many more individuals I wanted to talk to, as well as those of you reading this blog.
If you went to the event or were unable to attend but still have questions, feel free to contact me
See you next year!
Alexey Rybakov, interviewed by our staff blog writer Sarah Beaulieu