The biggest assumption enterprises make is that AI technology can solve all their business problems. In fact, most of the AI projects will fail and will not achieve its initial targets. I think to succeed with the AI project, the algorithm needs to access and process as much data as it can get, in order to train and fine-tune the algorithm. Without the right data, the AI algorithm will not be able to achieve its results.
How do you effectively utilise data while remaining conscious of privacy concerns?
There are solutions existed in the industry that allow you to utilise your data while preserving data privacy. Technologies like masking, tokenisation and encryption algorithms definitely can help. However, there are advanced use cases where those technologies will not be enough. For those use cases, technologies based on ‘hardware root of trust’, like Intel SGX or ARM TrustZone may help. This technology allows to process data in a Trusted Execution Environment (TEE), this TEE is isolated from the operating system and it guarantees that the code and the data uploaded into the TEE protected with respect to confidentially and integrity.
How has the industry’s attitude towards machine learning and other AI changed from when you first entered the field compared to today?
As I’m exploring solutions for enterprises to securely share data while preserving data privacy, I can see that there is a consensus on the fact that data needed to be shared, sometimes even between competitors, as long as the data will be used only for the agreed purpose. For example, think about a machine learning application for insurance fraud detection. If all insurance companies will have a way to share the data with their competitors only for this specific use case and the outcome will be the prevention of fraud in the insurance, I’m assured that all companies will participate. The issue is that no company can trust its competitor that the data will not be used for other purposes. But we’re very close to the solution that is based on TEE and in fact all cloud vendors are working on their offering for ‘confidential computing’ that leverages ‘hardware rooted trust’ in the cloud.
My main challenges this year will be to find the right partners to build the use case around this technology. Last year we focused on the technology and building the product as well as a couple of POC’s we did. Now we know what we can do and how we can ensure data security and data privacy for data analytics. The next challenge will be to find the right partners that have this concern.
In my presentation, I’ll talk about the problem of data sharing between parties for analytics and why so far there are not enough successes in this space. Besides, I’ll cover what are the existing solutions that enterprises can use and finally will share about a POC we did and how using TEE we can solve this problem.
Forward Marketing Summit Singapore 2020
18 Jun 2020 – 19 Jun 2020
The Forward Marketing Summit Singapore 2020 is going VIRTUAL. Faced with the unprecedented impact of the Covid-19 outbreak, we want to...Read More
Forward Marketing Summit Jakarta 2020
2 Sep 2020 – 3 Sep 2020
With Indonesia consisting of over 17,000 islands and some 700 languages, marketing in the country can present some of the greatest...Read More
Forward Marketing Summit Hong Kong 2020
2 Dec 2020 – 3 Dec 2020
Forward Marketing Summit is Hong Kong's leading conference where the greatest innovators of marketing, data and technology converge! Our...Read More