Machine learning is the ‘new normal’
- 16 April, 2018 06:45
Moving forward, it will be very natural to talk to computers
We are surrounded by machine learning, it is “under the hood” of systems we use everyday, says Olivier Klein, head of emerging technologies for Asia Pacific at Amazon Web Services.
These systems can range in the spam filter for emails, to the facial recognition systems used at airports.
“Cloud is normal, machine learning is the new normal,” says Klein.
Klein says machine learning helps organisations create new and seamless experiences for customers, or citizens, in the case of government agencies.
This can be through the use of a natural interface such as voice or facial recognition to create a frictionless experience for users.
“Moving forward, it will be very natural to talk to computer systems,” he says.
Klein, who spoke about the ubiquity of AI and machine learning at the AWS Summit in Sydney, says Amazon has been investing in machine learning for the past 20 years, using the technology in providing customer recommendations on its website, in Kindle, and in delivering packages using drones.
Klein says Amazon Echo and its voice controlled personal assistant service Alexa are examples of machine learning technologies that are now being used in business.
He says when Alexa was launched in Australia and New Zealand, a lot of businesses built their skills in order to use this as an extra channel for customers.
Alexa for Business, meanwhile, can allow users to use their natural voice to dial into meetings, schedule meeting and control equipment in the room such as air conditioners.
“For future businesses, machine learning will to a certain extent be a complete part of the business model," he says.
Collaborative robotics is everything
The key is to remove friction and to democratise machine learning models, says Klein.
“The goal of AWS is to put machine learning in the hands of every developer, every data scientist, every IT professional.”
They will have this fully managed machine learning services like Amazon Lex, Amazon Polly or Amazon Rekognition, and just straightaway use it, he says.
Experiment and build
Tim Dacombe-Bird, AWS New Zealand country manager, says trials on machine learning are happening across industries in New Zealand.
He says organisations can try a number of building blocks to get into machine learning.
“We have the tools available, they are inexpensive and they are available today.”
“The important thing is to be experimenting with it and trying things to understand how it can benefit the business,” says Dacombe-Bird.
“You need to define the problem that you are trying to solve with the technology. Once you have got the definition around that, then it is going to be a lot easier to solve these problems.”
The important thing is to be experimenting with it and trying things to understand how it can benefit the business
He says one of the things AWS provides to organisations is the ability to aggregate their data into a data lake.
This allows organisations to take advantage of analytics capability to think about, look at and determine what is likely to happen in the future.
“We are helping CIOs to think differently about their IT environments because the new machine learning components are really difficult to do when they are constrained in their own datacentres.”
“Whether it is proof of concept, or pilot phase or are now moving into production, we are helping them get real insights into their data.”
The new bot philosophy
Tye Brady, chief technologist of Amazon Robotics, says in the age of machine learning, “collaborative robotics is everything.”
He says humans have “amazing” capabilities such as being good at problem solving, abstraction, generalisation and creative thinking. Robots are good at crunching numbers, lifting heavy objects, moving with precisions and performing repetitive tasks.
If you can bring them together, “you gain a higher perspective.”
He says this is what happens at the fulfillment centres of Amazon. The inventory is held on pods, where the shelves are moved by robots. The operations pick products for orders, stow inventory to pods and take charge of inventory control.
“It is a massive parallel network, a symphony of humans and machines working together.
“That allows us to gain tremendous efficiencies inside our facilities.”
He says the new environment calls for new leadership principles, which he collectively calls his ‘bot philosophy’.
First is “customer obsession”. When you are customer obsessed, you start with the customer and work backwards, says Brady. The whole point of the use of robotic technology in the fulfillment centres is to delight the customer, he says.
Second, he says, is to “think big”. When you do this, you create and communicate a bold direction that delivers and inspires results. “Not only are we pioneers in robotics, we are helping shape what can be in robotics.”
When you are customer obsessed, you start with the customer and work backwards
Third, is to “invent and simplify”. We do this everyday, but this is not necessarily easy to do, he says. “It is a hard job to simplify but it is an important job,” he adds.
Fourth is to “learn and be curious”.
Be experimentalists and builders. “Never stop learning,” he says. “We are seeking ways to improve ourselves, we are trying new things all the time.”
Fifth, he says, is to “deliver results”
“When you do, you rise to the occasion.”
He says robotics systems have helped them produce their own delivery system, increase overall selection they offer to customers, and shorten delivery time that can take more than a day to less than an hour.
The author attended the AWS Summit in Sydney as a guest of AWS.