GSMA Global Mobile Radar: Spotlight on Artificial Intelligence
Published in September 2017
Questions from the webinar:
1) “It would be great to learn about some certain examples when telco operators use AI! Could you please give some?”
There a number of examples where AI is being used within operators already, and no doubt many more still to be uncovered. Popular examples of AI usage include network optimisation, fault prediction and proactive maintenance, service analysis and planning and of course, consumer behaviour prediction and analysis. It is also becoming more popular in technologies such as chat bots and service support ‘robots/agents’ where it can be offered to enterprise clients or used within operators own customer services environments. A limited number of operators are also launching consumer products with AI built in, NUGU from SK Telecom being a good example.
2) “Who truly leads the AI charge? Industry or academia?”
In pure, far out research, we would anticipate that academia would be more active, however as portions of this is industry-sponsored this position may not be so clear-cut. In truth though, we are seeing significant investment in both corporate, Government and risk capital funding research in AI and the vast majority of what we see in terms of output is being born out of these investments. So, from this perspective, i.e. practical and near future applications, we would think industry is taking the fore. It should also be noted, when it comes to large-scale social programmes such as smart cities, the role of Government should also be considered in this mix as it will have an important role to play.
3) “How will AI impact Software Development?”
AI is a cluster of technologies and as such has the power to transform a large number of markets and verticals, as a powering agent for all these, software and therefore software development will most definitely be impacted too. At its core, AI acts as a horizontal enabler, with solutions tailored to specific applications through the use of targeted teaching data that algorithms are exposed to. Software stacks will, therefore, need to adjust to interface with these enablers and tie them to tailored, possibly bespoke, data sets. The increased use of APIs and open development platforms, along with the importance of connected ecosystems, are therefore likely to play an increasing role within the world of software development. Security within software may also need to be rethought in this more layered, open ecosystem. Additionally, we are also starting to see examples of AI creating software itself; Bixby, Samsung’s AI voice assistant, and Microsoft’s Deepcoder, for example, are creating code themselves. Deepcoder uses a process called programme synthesis to create new code from existing software that it has been ‘trained on’. In short, AI has the potential to dramatically change both how software is structured and written to achieve and also who may be writing or optimizing it.