I wished this was my quote, but unfortunately, I’m not that famous 🙂
It’s a quote by the former head of Google Brain’s deep learning project, Andrew Ng.
Which I think it’s fascinating. Many people talk about “data as the new oil” and I agree.
Especially now, after Cambridge Analytica, everyone is aware of it.
But from my point of view, data is only the fuel of a bigger change which will shape a brand new world, lead by artificial intelligence (AI).
Everyone talks of AI, but what is it?
Everyone talks about AI but, what is it?
There are two types of Artificial Intelligence. The first, the Generic AI that is, a computer capable of doing anything a human can; and the Narrow AI, in which a computer does what a human can do, but only within narrow bounds.
We’re far away from the implementation of the Generic AI, but not very far from industries and jobs where basic tasks will be replaced by machines.
Think of self-driving cars, where machines will be able to drive people and goods from A to B more efficiently than humans. Picture yourself talking with a radiologist who has been working with an AI algorithm to recognise specific diseases and tumours.
In a world where many daily tasks will be performed by machines, humans will need to find a new balance.
AI and data, a very close connection
Now that we understood the role of AI, let me take back the initial concept. Why is data so important?
The digital revolution has produced a big amount of data. This data is extremely important for AI, as it’s the way machines are learning. Through data, machines are taught how a thing is done or what a thing is.
That’s why Google is giving us free space to store our pics and emails: to train its AI.
It’s fascinating. They basically act like kids, they learn from different data sources. The data for machine learning needs to be clean and vary to nurture the hungry AI.
Now, imagine: what would happen when all these inputs are based on a male-only analysis? AI is continuously learning thanks to data. If only men provide those inputs and interact with machines, our algorithm will continuously learn considering only one point of view.
Let’s consider some examples.
Early speech recognition machines were made by all-male teams, and they calibrated it for their voices. When they tried to sell to primarily female secretarial teams, the technology failed miserably.
A similar thing happened with car airbags. The team developing the airbag was all-male, and as they designed it, they used the height and weight chart for the standard man. The unintended and tragic consequence was that women and children were killed when those early airbags were deployed.
I don’t think this was the purpose. But it’s called unconscious bias because it’s the way some things are built by men considering their point of view only. It’s a weakness we need to solve together as we can’t count on an AI which is learning all the time from male-inputs only.
It doesn’t allow us to build an equalitarian future.
Artificial intelligence startups: who’s leading it NOW?
London is a key place for Artificial Intelligence startups. Silicon Valley is still the place where the majority of this research is. But China has recently taken the lead in this revolution.
You don’t expect it but China has become the home to many of the world’s top experts in AI and machine learning. SenseTime, a Chinese startup which makes AI-powered surveillance software for the country’s police, has recently become the world’s most valuable Artificial Intelligence startup. China has overtaken the US in terms of public and private investment in AI but, as you perhaps know, this is not the only metric to success.
The government is investing a lot of money in video surveillance in China. The future growth of this industry is a predictable trend.
Artificial intelligence startups are not only focused on self-driving cars and robots, then.
In fact, image analysis and recognition seem to have a big impact on artificial intelligence startups, too. And it’s not all. There’s another industry to focus on. Many experiments are nowadays around natural language processing. The interest towards understanding human voices and interaction is huge, especially for Google and Amazon. Amazon has Alexa and Echo on his hands with startups working on voice recognition games and apps for both.
Google has recently launched a couple of new experiments, too. The first is Talk to Books, where you can make a statement or ask a question, and the tool finds sentences in books that respond, with no dependence on keyword matching. This tool doesn’t work with complex sentences or questions, but it behaves well by answering raw factual questions. The second experiment is a game called Semantris, a word association game powered by machine learning. Curious? Play it, it’s free 😉
Many early-stage startups nowadays are looking to step into this sector developing apps based on natural language autonomous interactions.
But not that many of them are focusing on natural language processing, even if it’s the technology of the future.
Robots will need to talk to us. And whoever builds the best natural language interfaces today will dominate this industry.
How can artificial intelligence startups help marketing?
I’m particularly interested in AI and NPL tools because they can make easier and more exciting my marketing life.
I’ve already had the chance to work on Mixpanel and receive daily alerts on predictions, unexpected spikes and specific analysis from its AI algorithm. It’s particularly nice as you’re feeling helped on your daily data analysis by a sort of a “bigger eye”. You can dig into users preferences and habits with a huge degree.
I haven’t worked on e-commerce optimisation strategies, recently, but that’s where the semantic search can help, providing customers with the exact answers they’re looking for.
By 2020 it seems 85% of customer interactions will be managed without human intervention.
We don’t know yet how this will happen. But there’s only one question we can ask ourselves.
Are you ready for it?
[ITALIAN VERSION HERE]