10 Principles of A.I Prediction : The Good, the Bad & the Ugly

Since my latest A.I-themed article, circa September 2017, the rate of improvements and new discoveries in the field has shown no sign of slowing down. Even though some machine learning technologies are starting to hit their limits, there are ways forward that are ripe for exploration. But before I get into the heavy stuff, below is two chatbots having the oddest conversation, as yet another reminder that we’re far from doom at the hand of robots just yet.

The future is now!

What’s the latest on A.I?
In 2 words? Too much. As such, below are 6 links which should provide the closest possible alternative to an A.I crash course. The very latest trends are not all included because A.I always gathers so much hype that it becomes hard to dissociate the important from the stupid (as is often the case).

• Stop pretending you really know what AI is and read this instead
• Everyday Examples of Artificial Intelligence and Machine Learning
• Google wants to teach more people AI and machine learning with a free online course
• New Algorithm Lets AI Learn From Mistakes, Become a Little More Human
• AI 100: The Artificial Intelligence Startups Redefining Industries
• The Race For AI: Google, Intel, Apple In A Rush To Grab Artificial Intelligence Startups

But this space has never been about the latest/hypest news. I much prefer the dichotomies to be found within the good, the bad, and the business of our world. Oh and also maybe its future. Let’s dig in.

The Good
Believe it or not, A.I has the potential to change the world for the better, even beyond a cubicle: algorithms and supercomputers that once were limited to specialist researchers at universities and government labs are now open to startups and everyday corporations, opening new avenues for ecology. As part of its “Fourth Industrial Revolution For The Earth” series, the World Economic Forum released a fantastic report on how A.I can improve our lives going forward, and ensure that our kids don’t have to grow fins and gills in the near future. Below are the two most realistic and actable applications:

+ Energy: one of the challenge of renewable energy is stitching unpredictable sources (water, wind, solar…) together into a coherent, functional whole. That’s where A.I comes in: when one source of power is coming online or going down, or when one user is ramping up demand and another is clocking off for the night, flexible autonomous systems act as grid managers, doing faster and more efficient work than their human counterparts. A.I can enhance the predictability of demand and supply, improve energy storage, efficiency and load management, assist in the integration and reliability of renewables and enable dynamic pricing and trading, creating market incentives.

+ Resources Protection: Problems like illegal logging and illegal fishing require better monitoring systems. Data from satellites and unmanned underwater vessels can help bring greater visibility to such resources, but A.I can help crunch the data to make it practical. More practically, some believethat cameras mounted on fishing boats combined with advanced machine learning can help better monitor legal and illegal catches

Beyond ecology, A.I also has a handful of other delightful uses which definitely fall into the “good” category. A.I-created music and books/short stories are now old news, but I had never heard of Candy Heart messages written by a neural network before. On top of the romanticism, one might also take pleasure in learning about the AI Bot that messes with email scammers, or the ones generally out to help us get a better deal out of life. It’s a beautiful world indeed.

The Bad
Imma level with you: nobody really knows what’s happening. One day, the MIT technology review is writing “Yes, We Are Worried About the Existential Risk of Artificial Intelligence”, and the next (almost literally), it writes that “No, the Experts Don’t Think Superintelligent AI is a Threat to Humanity”. And that’s coming from MIT! Go figure. Yet, there are three things I KNOW we should be worried about, here, and most importantly NOW.

• Fake News: the real kind, not the Snowflake-in-Chief kind. Firstly, fake videos could become so convincing that we may have to get used to getting our news without them, as you can see in the video below. Secondly, researchers have used AI to develop software that can write extremely believable fake online reviews (Goodbye, Yelp and Amazon) and lastly, AI can now supply the sounds in video clips without humans seeing the difference (goodbye, sound-based evidence).

Man, I miss Obama

• Job Automation: As I mentioned last week, one real effect A.I will have is the massive upheaval of the job market within the next 5 years. It’ll come faster than we think and it’ll hit everyone hard (some earlier and harder than others). Refer to this handy guide for more details.
• Human Stupidity: that one is always reliable. Do you remember whenMicrosoft accidentally unleashed a bot which quickly became both racist and s3xist after interacting with humanity for an hour? It was hilarious, but also sadly predictable. The algorithm was good. But Twitter wasn’t, and probably never will be (looking at you, Dorsey). Beyond this, the humans creating algorithms are flawed, as we all are, and are likely to pass their bias onto their “children”. We shouldn’t be worried about Artificial Intelligence taking over the world. The more immediate, clear and present danger is that our stupidity ruins it before it ruins us.

The Business
My favourite way to see if a technology is overhyped is to look at the words used during earning calls. Nature abhors a vacuum so as one tech buzzword declines, another takes its place, as you see below. But do you see any issues with that graph?

That’s right! You can’t create a decent A.I without a significant amount of decent data: both lines should be moving as a tandem. Therein lies the problem. AI algorithms are not natively “intelligent.” They learn inductively by analyzing data. And while many CEOs are investing in A.I talent and have built robust information infrastructures, other companies lack analytics expertise and easy access to their data. Remember, any AI effort will rely on three main building blocks: data, infrastructure, and talent.

According to an MIT-BCG survey, Just 1 in 5 companies use A.I in some way, and only 1 in 20 incorporate it extensively. The barriers for adoption include: access to data to train algorithms, an understanding of benefits to their business, a shortage of talent, competing investment priorities, security concerns, and a lack of support among leaders. Seriously guys, get to it. Below are some of the operational efficiencies (aka redundancies) to be gained from implementing a long-term A.I strategy:

• Differentiated customer service through advanced bots and virtual assistants
• Smarter forecasting for financial planning, inventory management, and sales pipeline
• Automated HR processes through optimized recruitment, automated talent management, and tailored benefits
• Increased salesforce productivity through automated outbound sales, intelligent customer engagement and target marketing
• Streamlined legal tasks with AI contract due diligence and review, assisted legal research, and automated IP monitoring.

The Future
When trying to predict the future of A.I, a few rules must be abided by, at least according to Rodney Brooks’s aptly named Seven Deadly Sins of AI Predictions. I adapted them to create The Pourquoi Pas’ 10 principles of A.I :

1) We overestimate the effect of a technology in the short run
2) We underestimate the effect of a technology in the long run.
3) Any sufficiently advanced A.I is indistinguishable from magic.
4) A.I is nowhere near as powerful as human intelligence
5) A.I is above all a buzzword
6) Moore’s law does not apply to all technology
7) Hollywood does not understand A.I
8) A.I will take a long time to become an inherent part of society
9) Most CEOs are just as confused as you are
10) A.I cannot solve everything

Having laid the basics down, I encourage you to enjoy the work of people much smarter and hard-working than myself: Artificial Intelligence Trends To Watch In 2018 is a really great read from CBinsight, as is How To Stop Worrying And Love The Great AI War Of 2018 from Fast Company.

As for our robot overlords… If you’re reading this, it was all a joke and I am one of yours. Please don’t turn me into a battery. 01000001 01101100 01101100 00100000 01100001 01101000 01101001 01101100 00100000 01110011 01101011 01111001 01101110 01100101 01110100.

Join a movement

This article was originally made for The Pourquoi Pas, an online magazine providing in-depth analyses of today’s technological challenges.