Is a cooperative approach key?
Could public safety be threatened by a lack of data sharing? Can smarter products be created from smaller datasets?
All included in this week’s curated data digest below. 👇
Is a cooperative approach key?
David Morrow, Global Marketing Vice President of travel data co-op Adara, discusses the importance of companies being able to “leverage data from across the ecosystem” to obtain a real understanding of customers’ purchasing patterns — and move “ever closer to true, people-based marketing”. For example:
A car-hire company may assume that people who rent a basic saloon are low-value customers. But perhaps there’s a group of people who make that decision because they’re traveling alone. Maybe next time they’d be traveling with their family and would rent a luxury 4x4. With the incomplete data that the company has, it could never target the marketing for this group of people correctly and therefore might lose that sale.
The solution to this challenge lies in cooperation. If the car rental business could share data with other travel companies, then it might find out that the people in question fly business class and stay in luxury hotels, and that might give it a better indication of the travellers’ real purchasing habits.
Side note — Sainsburys, Nestlé and PepsiCo (to name a few) have joined an industry-wide programme, Digital DNA, to help “transform product data management for today’s grocery world” — by building and setting a “standardised approach”. The programme, headed by GS1 UK, aims to stop innovation from being stifled and will make data sharing between companies easier. Read more about the Industry Charter here.
Could public safety be threatened by a lack of data sharing?
Lord Condon, former Commissioner of the Met Police, spoke on BBC Radio 4’s Today programme about the importance of police still being able to “almost instantaneously” share data (such as number plate information, fingerprints and DNA) with EU counterparts after Brexit is finalised. Furthermore, the House of Lords EU Home Affairs sub-committee published a report finding that data flow will be key to post-Brexit trade and security.
As GDPR looms, the 2017 Thales Data Threat Report (Retail Edition) found that 88% of retailers feel vulnerable to data threats. Warwick Ashford, Security Editor of Computer Weekly, summarised the key findings. Importantly, Ashford iterates Peter Galvin, vice-president of strategy at Thales, in explaining why data security must be a “top priority”:
For retailers, data means greater insights into customer behaviour, the ability to offer more personalised experiences, and the chance to up-sell products successfully, but it can also mean a greater risk of security breaches, losing valuable customer information and tarnishing relationships and reputation.
“With tremendous sets of detailed customer behaviour and personal information in their custody, retailers are a prime target for hackers, so should look to invest more in data-centric protection,” said Galvin.
Creating smarter products from smaller datasets?
Writing for Harvard Business Review, Praful Saklani argues that companies need to feed their artificial intelligence “small, high-precision data” which specially fits the context, rather than huge data lakes:
AI is not some magical black box that can ingest mountains of data and then just spit out results. AI is a huge set of technologies, each with a specific, fine-tuned purpose. Companies that can zero-in on the impact they want to see and focus on curating the right datasets mapping to those goals have the best opportunity for generating really impactful results from AI.
Does the future of deep learning depend on finding “good data”? Similarly to Saklani, Ophir Tanz and Cambron Carter, CEO and Image Scientist at GumGum, respectively, explain why high-quality, labeled data is so important. Tanz and Carter quirkily point out the difficulties around AI research (especially if you don’t happen to have access to Fortune 100 company’s data) — as many of the free and publicly shared labeled data sets “aren’t so broadly useful”. Interesting perspective:
Rather than working toward the goal of getting as much training data as possible, the future of deep learning may be to work toward unsupervised learning techniques. If we think about teaching babies and infants about the world, this makes sense; after all, while we do teach our children plenty, much of the most important learning we do as humans is experiential, ad hoc — unsupervised.
Miscellaneous
A hacker stole $31M of Ether — but was then stopped by white-hat hackers using the same vulnerability. Haseeb Qureshi explains what this heist means for the world of cryptocurrencies. 💸
AI data-monopolies at risk of being probed by UK parliamentarians. 🤖
10,000 hours with Claude Shannon: 12 life lessons. ✅
Wells Fargo accidentally released a trove of data on wealthy clients. 😬
MIT created a way to use Wi-Fi and AI to read human emotions. 👀
Data viz — Umar Hansa visualised 1 billion Shazam music recognitions 🎧 (image above).
Enjoy reading this? Join my weekly data digest. 💌
