Comment: How AI and convergence turns electronic engineers into invention developers
- Auteur:Ella Cai
- Relâchez le:2018-08-15
Julian Nolan, CEO of data-driven invention technology company Iprova, explains how AI could affect electronics and how electronic engineers can take advantage to help deliver innovation.
If there is one thing to be learned from the numerous consumer electronics trade shows that have taken place in recent years, it’s that leveraging artificial intelligence (AI) in electronic devices is high on the agenda.
You only need to look at the keynotes of this year’s CES or upcoming IFA – AI is everywhere. Yet, the link between AI and the consumer devices that are taking centre stage at these shows is still ambiguous.
For better or worse, the Consumer Electronics Show (CES) 2018 was emblematic of where the electronics industry currently sits in its adoption of AI. While, on one hand, it showcased that embedded AI chips are becoming widespread and the number of players in that market is set to increase, it also showed that many electronic device manufacturers conflate AI and smart functionality.
Simply put, an algorithm is not necessarily AI. Most smart devices use algorithms that follow relatively straightforward logical processes.
For example, a smart microphone could have embedded algorithms that normalise input audio levels and clean out an adjustable level of background noise. If it does this in real-time to minimise the need for post-production, it doesn’t mean the microphone uses AI; it is using algorithms that are able to process data very quickly. It’s impressive, but it isn’t quite AI.
Smart is not always AI
This misconception of AI in electronics has also been identified by market analysis company Futuresource. Simon Bryant, the company’s associate director of consumer electronics, has stated that, “devices may well be smart or intelligent but that doesn’t mean they are AI. The intelligence could be algorithms or logic to improve performance or data processing, and analytics to detect trends and provide information back to the user, but it’s still not AI.”
Currently, there are few electronic devices on the market that make use of AI. This has been a topic of discussion in the industry for the past year, with some suggesting that the next step forward for certain devices, like drones, is to incorporate machine learning algorithms into them for more intelligent, autonomous operation.
Achieving this will require further development, not least in deploying more advanced machine learning models onto smaller objects. Researchers from Microsoft in the US and India are currently looking into compression techniques such as weight quantisation and sparsification to reduce the footprint of neural networks for more compact devices.
As of June 2017, these compression techniques allowed for a 10–100x reduction in size. Further size reductions don’t seem to be attainable without sacrificing accuracy, so the team are also working from the ground up to develop neural networks 1,000–10,000 times smaller without the performance trade-off.
But while we’re still years away from truly using AI in many electronic products, AI is already beginning to transform the development of new electronics at the invention stage. Iprova for example uses advanced machine learning to converge myriads of data from seemingly unrelated industries to generate truly novel inventions for their clients.
Taking advantage of industry convergence
The convergence of previously isolated industries is a phenomenon that provides unprecedented opportunity for agile businesses, and an operational threat to organisations that are slow to adapt.
In 2014, a Gartner report concluded that convergence represented, “the most fundamental growth opportunity for organizations [sic] and will redefine industry boundaries by shifting the focus from individual products to cross-industry value experiences”. Two years later, an IBM study identified that many C-suite executives were concerned about adjacent players entering their market space due to industry convergence.
However, spotting opportunities for invention in convergent markets is increasingly difficult due to ever growing amounts of data. As former Google CEO Eric Schmidt famously stated, “between the birth of the world and 2003, there were five exabytes of information created. We now create five exabytes every two days”.
Included among this data are new pieces of research that could show an opportunity for convergence between two industries. The trouble is, using that data effectively to create the next generation of electronic devices is no easy feat. The human ability to review information is intrinsically limited in terms of breadth and speed.
Machine learning
This is where machine learning comes in. It analyses data in real-time and identifies the inventive signal among huge amounts of information noise. Iprova is at the forefront of this digitalisation of invention. The company uses its advanced machine learning technology to augment the human ability to invent new products and services, accelerating the invention process and producing truly novel results.
This is particularly important for companies that want to add new value to their electronic devices, because speed and invention diversity is key for bringing new disruptive products and services to converging markets at the optimum time.
Despite the electronics industry operating so much more quickly than ever before, the world will never again change as slowly as it does today. There is an increasing need to invent ever faster, and with ever greater levels of disruption. Iprova’s AI technology helps to create inventions in response to real-time technical, social and market advances, meeting the challenges posed by the ever-increasing rate of change.
From electronic engineer to invention developer
Electronic engineers are in an interesting position with the use of AI in product development; they are essential in realising the convergent future of invention. In most scientific and engineering disciplines, there has been a trend towards ever increasing specialisation. However, converging industries and ever faster rates of change demand a different set of skills.
Whilst there is still a need for deep and fundamental knowledge of science and technology, inventing in convergent markets requires a broader skill set that can be widely applied over many physical sciences areas, from anatomical intelligence through to autonomous vehicles. In other words, this skill set is what turns electronic engineers into invention developers. By leveraging market convergence and using a data-driven approach to invention such as Iprova’s, invention developers are uniquely positioned to to create disruptive, timely inventions that have a significant impact on tomorrow’s products and services.
As a result, while consumer electronics events show that there is a lot of mystery surrounding the use of AI in the electronics industry, there’s an opportunity for a new breed of electronic engineers to leverage machine learning technology. We might be a few years away from seeing genuine AI embedded into devices, but using AI at the invention stage is here today and already helping to define the products and services under development by some of the world’s most respected technology companies.
If there is one thing to be learned from the numerous consumer electronics trade shows that have taken place in recent years, it’s that leveraging artificial intelligence (AI) in electronic devices is high on the agenda.
You only need to look at the keynotes of this year’s CES or upcoming IFA – AI is everywhere. Yet, the link between AI and the consumer devices that are taking centre stage at these shows is still ambiguous.
For better or worse, the Consumer Electronics Show (CES) 2018 was emblematic of where the electronics industry currently sits in its adoption of AI. While, on one hand, it showcased that embedded AI chips are becoming widespread and the number of players in that market is set to increase, it also showed that many electronic device manufacturers conflate AI and smart functionality.
Simply put, an algorithm is not necessarily AI. Most smart devices use algorithms that follow relatively straightforward logical processes.
For example, a smart microphone could have embedded algorithms that normalise input audio levels and clean out an adjustable level of background noise. If it does this in real-time to minimise the need for post-production, it doesn’t mean the microphone uses AI; it is using algorithms that are able to process data very quickly. It’s impressive, but it isn’t quite AI.
Smart is not always AI
This misconception of AI in electronics has also been identified by market analysis company Futuresource. Simon Bryant, the company’s associate director of consumer electronics, has stated that, “devices may well be smart or intelligent but that doesn’t mean they are AI. The intelligence could be algorithms or logic to improve performance or data processing, and analytics to detect trends and provide information back to the user, but it’s still not AI.”
Currently, there are few electronic devices on the market that make use of AI. This has been a topic of discussion in the industry for the past year, with some suggesting that the next step forward for certain devices, like drones, is to incorporate machine learning algorithms into them for more intelligent, autonomous operation.
Achieving this will require further development, not least in deploying more advanced machine learning models onto smaller objects. Researchers from Microsoft in the US and India are currently looking into compression techniques such as weight quantisation and sparsification to reduce the footprint of neural networks for more compact devices.
As of June 2017, these compression techniques allowed for a 10–100x reduction in size. Further size reductions don’t seem to be attainable without sacrificing accuracy, so the team are also working from the ground up to develop neural networks 1,000–10,000 times smaller without the performance trade-off.
But while we’re still years away from truly using AI in many electronic products, AI is already beginning to transform the development of new electronics at the invention stage. Iprova for example uses advanced machine learning to converge myriads of data from seemingly unrelated industries to generate truly novel inventions for their clients.
Taking advantage of industry convergence
The convergence of previously isolated industries is a phenomenon that provides unprecedented opportunity for agile businesses, and an operational threat to organisations that are slow to adapt.
In 2014, a Gartner report concluded that convergence represented, “the most fundamental growth opportunity for organizations [sic] and will redefine industry boundaries by shifting the focus from individual products to cross-industry value experiences”. Two years later, an IBM study identified that many C-suite executives were concerned about adjacent players entering their market space due to industry convergence.
However, spotting opportunities for invention in convergent markets is increasingly difficult due to ever growing amounts of data. As former Google CEO Eric Schmidt famously stated, “between the birth of the world and 2003, there were five exabytes of information created. We now create five exabytes every two days”.
Included among this data are new pieces of research that could show an opportunity for convergence between two industries. The trouble is, using that data effectively to create the next generation of electronic devices is no easy feat. The human ability to review information is intrinsically limited in terms of breadth and speed.
Machine learning
This is where machine learning comes in. It analyses data in real-time and identifies the inventive signal among huge amounts of information noise. Iprova is at the forefront of this digitalisation of invention. The company uses its advanced machine learning technology to augment the human ability to invent new products and services, accelerating the invention process and producing truly novel results.
This is particularly important for companies that want to add new value to their electronic devices, because speed and invention diversity is key for bringing new disruptive products and services to converging markets at the optimum time.
Despite the electronics industry operating so much more quickly than ever before, the world will never again change as slowly as it does today. There is an increasing need to invent ever faster, and with ever greater levels of disruption. Iprova’s AI technology helps to create inventions in response to real-time technical, social and market advances, meeting the challenges posed by the ever-increasing rate of change.
From electronic engineer to invention developer
Electronic engineers are in an interesting position with the use of AI in product development; they are essential in realising the convergent future of invention. In most scientific and engineering disciplines, there has been a trend towards ever increasing specialisation. However, converging industries and ever faster rates of change demand a different set of skills.
Whilst there is still a need for deep and fundamental knowledge of science and technology, inventing in convergent markets requires a broader skill set that can be widely applied over many physical sciences areas, from anatomical intelligence through to autonomous vehicles. In other words, this skill set is what turns electronic engineers into invention developers. By leveraging market convergence and using a data-driven approach to invention such as Iprova’s, invention developers are uniquely positioned to to create disruptive, timely inventions that have a significant impact on tomorrow’s products and services.
As a result, while consumer electronics events show that there is a lot of mystery surrounding the use of AI in the electronics industry, there’s an opportunity for a new breed of electronic engineers to leverage machine learning technology. We might be a few years away from seeing genuine AI embedded into devices, but using AI at the invention stage is here today and already helping to define the products and services under development by some of the world’s most respected technology companies.
