Quantcast CTO and machine learning expert Peter Day cuts through the adtech jargon to provide a glimpse into the future.
Machine learning is not the same as artificial intelligence (AI).
Both terms are in hype cycles but they’re actually a lot simpler than made out to be.
AI is an umbrella term, simply describing machines that demonstrate ‘intelligence’. It’s a vague concept that lacks a rigorous definition. To start, how does one even define human intelligence?
Machine learning on the other hand is a formal discipline. At its core, it asks ‘how can machines solve problems without being programmed explicitly by humans?’
There are biases in AI but, contrary to popular belief, it’s not because of the programmer.
Bias is another term often thrown around without a proper understanding of its cause. Making sense of the world requires a certain level of generalisation or bias. The human brain is constantly taking into account millions of signals a second, but your conscious mind only receives a few of them. To work out which signals to ignore and which to highlight, your brain is constantly making generalisations and assumptions — leading to a biased view of the real world.
When we apply techniques like machine learning, the same principle applies.
Machines find patterns in observations of things that have already happened (data) in order to make inferences and decisions. This opens the results up to bias.
These biases can come from a couple of root causes. The first is in the data collection. For example, if more data is collected on dogs rather than cats, the machine will be better at detecting dogs. But machines also find patterns, reflecting biases that unfortunately exist in the real world. For example, when Amazon attempted to use machine learning for recruitment, it encountered a serious problem when it filtered out female resumes based on a long history of hiring men.
Adtech companies do care about privacy.
At Quantcast we take privacy extremely seriously. Privacy-by-design is in our DNA. We put a lot of effort into partnering across the industry to put consumer privacy first while helping advertisers and publishers achieve their business goals
For example, we partnered with the IAB and other actors across the ecosystem to create the transparency and consent framework (TCF). We also created an easy-to-adopt implementation of this, Quantcast Choice. Launched in 2018, Choice is now the number one consent platform in the US and Europe, giving internet users the choice of who they want to share their data with.
Machines aren’t evil but they will change the future of work.
Like most new technologies, machine learning is disruptive. But there’s time to adapt. Unlike the prophesied sci-fi robot takeover, realistically we will see change in basic operational tasks where the gap between idea and execution is long. For example, transferring information from emails to spreadsheets can be automated, freeing up time to try new creative ideas at a greater scale.
I’m optimistic about the future. We’re going to see disruption across all industries, particularly in areas where if mistakes are made, the damage is limited. Think supply chain optimisation, retail, transport, and more disruption within our own industry.
Have something to say on this? Share your views in the comments section below. Or if you have a news story or tip-off, drop us a line at adnews@yaffa.com.au
Sign up to the AdNews newsletter, like us on Facebook or follow us on Twitter for breaking stories and campaigns throughout the day.