Machine Learning

A.I. in Advertising – No Longer Hype
The AI Hype Cycle

The Technology Hype Cycle: When does technology really become usable.

A.I. is nowhere near where we all want it to be. Yes, there are lots of breakthroughs happening, but we simply just don’t yet have the technology to, for example, converse freely with a computer. Or as Earthling founder Caleb Eastman explains “don’t drop a Tesla on the Moon and expect it to explore and learn.”

Areas where we are seeing real-world examples of computers helping us think in real-time generally include scenarios where there is easy access to large sets of data. This is when it starts to become useful.

Products like Grammarly for example not only are able to be somewhat easily trained due to the fact that there are mass repositories of human language, plus ongoing personal users’ language to use as a reference.  This is, in essence, is much easier to solve programmatically than trying to recognize anything anyone says via voice for example.

Another area where Artificial Intelligence is already helping – advertising. Modern advertising is a function of data. Intelligent engines abound such as Finch or Perfect Audience  which use intelligence to try to optimize advertisers’s budgets, retargeting and conversion.

This gives the advertisers a huge advantage online, a medium plagued by bots and battling algorithms and only truly leveraged by those using automated software.

The hype is real and ultimately all of these intelligent devices will become a reality. But it takes time for industries to mature and for humankind to realize exactly how to leverage new widgets. The “A.I Era” is unique in that technology needs lots of data to learn how to be intelligent. So naturally, those industries with access to and the ability to manage large amounts of data will benefit first from this technological revolution.

Stay tuned, it’s an exciting time.

December 28, 2019 by Artificial Intelligence Machine Learning

A starting point for Machine Learning: Stanford’s Andrew NG

Take a breath. Machine Learning is really nothing new. Remember stats class?

It does take some education. Computers can do much of the work these days. But you as a data scientist need to understand what you are doing, how to manage and classify data and which questions to ask. Like any other career, it takes experience.

Andrew NG ‘s Stanford Machine Learning class is free, easy to follow and considered the gold standard for Machine Learning.

You can watch the first class here:

Adjunct Professor and Coursera Co-founder Andrew NG teaches Machine Learning at Stanford
December 22, 2019 by Machine Learning

We’re all in on the Wildfire XPrize
We are seeking collaborators for the upcoming XPrize designed to drive innovation around fire detection and prevention.

The Roaring ’20s are Here

A new decade is here. The “Roaring ’20s”. It almost feels like a new era emerging!

One of the things we wanted to do at Robauto is set some moonshot goals for the upcoming years. We want to put to work our platform and years of experience launching community driven technology projects.  Our expertise is in hardware and software. Robotics, Machine Learning and IoT Networks as well as Sensors and Drones are what we can contribute.

Let’s Figure out a Smart Way to Detect and Extinguish Wildfires

Unfortunately one of the things likely to be roaring during the ’20s is our forests. As climates change and human populations expand, wildfire is a growing problem. This is an important issue for everyone. Being a Colorado company, Robauto has even more incentive to help protect and preserve our natural forests. We also think it’s a perfect example of how technology and people can work together to help their communities in a good, non-intrusive way. 

Machine learning, community involvement and IoT networks can help. In fact, detecting, preventing and managing wildfire is the perfect application of sensors, data, intelligence, and automation. We are early in our product development process and need input from fire experts, community members and engineering collaborators. 

Collaborators Wanted: Wildfire XPrize Project

We need your help. We’ve come up with a few general principles we want to adhere to but otherwise are in the information gathering phase. 

  • The solution should involve technology, the natural environment and people.
  • The solution should be non-invasive. We want to keep our wild spaces wild. This includes limiting signals, hardware, robots and people where possible.
  • The solution should detect, contain and assist with extinguishing any fire in any location on the globe.

Interested in participating? Email support@robauto.ai. 

 

December 5, 2019 by Artificial Intelligence Internet of Things Machine Learning

Data is the new oil

Economists have been claiming for a while that data is now a more valuable commodity than oil. I first heard this quote during a Netflix movie (The Great Hack) about the Cambridge Analytica role in the 2016 presidential elections. Regardless of what you think about the ethics of it all, it raised an interesting point about how valuable data has become. Not just our personal data but all forms of it.

As someone who has built data-driven startups (Infopia, Yovia, MEC Labs, Robauto) I can testify first-hand that data is an incredibly valuable commodity. Whether or not it’s more or less valuable than oil – I can’t really say. They don’t totally compare.

What I can say, however, is that data is the key to artificial intelligence. A common dialogue around the data discussion has to do with the privacy and tracking of consumers. Quickly, people start looking at companies like Google and Facebook, who seemingly track and create experiences for us based on our behavior. The answer is that they do 100% track our behavior and use it to optimize their revenues and provide a more tailored experience. Your like, share and search data trains their software to give you a better experience.

This is nothing new. Supermarkets have been tracking your behavior for years. Loyalty programs tie your in-store behavior to purchases to maximize profits. This isn’t some covert attempt to learn more about us – this is their job. A retail store’s goal is to make money and it’s really useful to look at data and adjust. Computers use data sets to try to pre-train themselves so more data is better. In the case of the Netflix show the premise was that it was a system of posts, social events, and even physical groups that were perpetrated by targeting a subset of the population that the algorithm had shown as being easily to influence.

The data is what drives machine learning. Without data, underlying algorithms don’t work. They need lots and lots of data. From robots to IoT devices to web advertising, data is what feeds the proverbial machines.

The example of Cambridge Analytica using data is extreme. In the Netflix documentary, they supposedly had a weapons-grade software algorithm that used questionably obtained Facebook user data to create more than 5,000 data points around every voter in America. They then used that data to figure out how to influence people who were undecided in the 2016 elections by sending them targeted candidate propaganda. I  personally find that a little invasive and not at all transparent but it’s not new, particularly in advertising.

We can argue whether or not it is ethical or if it is more or less valuable than something like oil, but the fact of the matter is this:

With a sample of less than 0.06% of the world’s Facebook data, Cambridge Analytica was able to help sway an election. That’s all it took for their algorithm to identify who to reach and what to say to them to influence their vote.

The show depicts their efforts coming into the public spotlight when the Trump campaign won in an upset and it was revealed that this organization had been swinging elections for years. It was their business, and they were good at it. Perhaps they went too far and I can’t really comment on the ethics of it other than I would never want to use technology for something so…? and I don’t know people who do. It seems like an unhelpful way to use data but it illustrates the point of its potential value.

Tesla is an example of a company using data. You can say what you want about their products or the company or the stock price, but Tesla has a huge advantage in that their underlying artificial intelligence – the self-driving software – has already been fed millions of miles of live driving data.

This gives the algorithm an advantage over someone who maybe has a great vehicle and software platform – but no data to train it.

Nobody really knows what the future holds for technology, but one thing for sure – your own personal data may become one of your most valuable assets. Ironically, your human input is what brings these technologies to life and for the most part, we all willingly feed it.

Remember Stats? Machine Learning is set to take over.
July 30, 2019 by Artificial Intelligence Machine Learning

What is Artificial Intelligence (A.I.)?

What do you think of when you hear the words Artificial Intelligence?

For me as a kid it was robots. I imagined a future world filled with walking, talking robot companions. Robots to carry me to school. Robots to sit and play a game with me. You may remember The Jetsons. It was a cartoon  which depicted a nice family who lived in a futuristic world of flying cars and friendly personal butler robots. I dreamed of one day living in a world like that.

And today flying cars are real and personal robots exist by the thousands. But I still mop my own floor and drive a pickup truck. Like most fiction it’s not exactly how the creators of the Jetson’s imagined.

While most of the technology seen in this futuristic show exists – it’s evolving a little differently than we thought.  And It’s also nothing to joke about. AI is more than just robots. It’s software that is already changing the world. AI is not as complicated as is sounds and anyone can become an AI innovator.

What is Artificial Intelligence (AI)

WikiPedia defines A.I. a computers demonstrating intelligence. So in simple terms the computer is making some sort of decision based data it’s fed.

But all of this is nothing new. There’s been lots of work done at government and University levels research facilities since the 1950’s. The Ancient Greeks conceived a similar idea and semi-intelligent software has been around for a long time. So what is all the buzz? For starters A.I. has simply become a trendy buzzword. I have seen a number of startups suddenly emerge that claim use some semblance of AI in their product. Some of these are nonsense and a few are likely very good and will become useful.

In reality it is hard to find anyone with any kind of new truly functional, useful AI like is depicted in movies. This is just because the technology isn’t quite there yet. Even though we have lightning fast computers and Internet available, it’s not always cost-effective or feasible to process mass amounts of data in real time.

This is why you see prototypes with patches of intelligence displayed in closed environments but not as many fully functional human looking robots out there ready to greet you throughout your day.

Getting started on some basics: There are 2 main types of AI:

Applied: Most common – a very specific application such as a self-driving car or a voice activated computer. These are using libraries of data and get better as they go. These software programs typically use some sort of Machine Learning architecture where the engineer uses lots of machines each receiving inputs on the environment. In aggregate, or over time, the machines are able to find a pattern and use the pattern to make decisions. 

General: I can’t think of a single ‘general’ AI that actually works in a production mode. If someone has one please let me know! An example would be a robot  you could walk up to and it would just instantly be able to converse freely with you with no human input. Even the impressive humanoid Sophia isn’t really fully autonomous. 

AI as a socioeconomic tool

Over the years I’ve had the opportunity to host and participate many robotics meetups and events with thousands of people and virtually every type of robot known to man. I’ve designed and brought to market several technologies, some of which failed and a few which made it through the gauntlet of consumer acceptance and adoption.  For a working class kid from rural Vermont technology was an opportunity to transform myself and the world around me.  Today I love my job helping others to innovate around robotics.

I am a capitalist so yes I want to also make a profit. That’s important to sustain innovation. But teaching people about AI, robotics and technology entrepreneurship is my social-economic-spiritual statement to the world, and gives me a purpose. Currently I’m working on BiBli which is a robotics platform to help people learn about all of this easily and inexpensively. It’s also really fun to collaborate with smart, talented people to make goofy robot inventions and even some potential breakthroughs. 

The CEO of Google recently came out and said he thought AI was more of a game changer than electricity. Elon Musk has warned AI is more dangerous than nuclear weapons. I agree with both statements. Do we need to be afraid? No. Do we need to pay attention? Yes.

AI is here and already impacts your life daily.  Marketing, fake news, hacking, the financial markets, security and safety screening, healthcare and education all use AI.

It’s really nothing new, nor that difficult to understand conceptually. If you are a real geek  interested in the math and more advanced aspects of AI I would suggest starting with learning a about the basics of neural networks and or the different types of machine learning.  Being versed in college-level math such as calculus or linear algebra could help you to understand but really all of this comes down to analyzing data. 

If you have no programming experience, don’t worry that part isn’t important in the conceptual stages. If you are really interested you could spend a few nights a week learning Python which is a great starting point to learn software.

Simply start to think about how a robot could help you in your daily life

During Build-a-BiBli workshops we often give people a blank piece of paper and ask them to design their ‘dream robot’. What comes back is pretty typical. In the case of students their robot usually looks like a robot they’ve seen in movies, cleans their room and does their homework.

We talk about the power usage required to power a laundry folding arm and how difficult it would be to program a computer to do homework in any language on any subject. Eventually the robot design gets reduced down to a prototype idea that could actually be built and used.

An interesting note – In the case of adults there is almost always one person who wants a robot that will go get them a beer. In all cases, their robot is usually inferior – a sort of slave designed to serve them.

An Example of an AI Project in Real Life

Sometimes the easiest way to understand a concept is to see it in real life. For this example let’s take a look at Tesla’s self-driving cars. You may own a Tesla or hopefully have at least seen one drive by. Tesla isn’t the only autonomous driving vehicle out there but they are pioneers.


It’s  a great example of a network of computers working together to get smarter. Teslas rely on cameras, sensors, GPS data and a driver to navigate through streets and highways. And while the self-driving is getting better and better it’s still not totally autonomous. The company has given no deadline for when they will be.

Think about why:

There are many Teslas in the world and it was recently reported that more than 1.3 Billion miles have been driven autonomously. They are all connected to the internet. Data from all of those miles have been fed into the Tesla software. Also the data from your own routes as a driver  (hint: you repeat the same ones over and over) are also part of the math.

I would guess that when there is enough collective data to make that system perfect they will offer a software update that makes of the cars truly intelligent and self-driving will be a reality in the world. And they will likely make a pile of money in the process.

The Tesla fleet is a perfect example of neural network using machine learning to get smarter as they go. This is a real thing – in fact you can even see discussion about the Tesla neural network community here.

Get started inventing the next AI

There’s no calculus homework involved. You won’t need a soldering gun or engineering background to complete this challenge.  You just need creativity:

    • Today, simply go about your day. Look at all of the devices in your life. Appliances, computers and vehicles are everywhere. Most are connected already to  or bluetooth. Think about which of those could help you more in life if they were just slightly ‘smarter’.
    • Then think about the types of data they could potentially collect if they all had inputs or sensors. Sensors are cheap and easy and just send a computer a signal that is either on, off or with some value. They can detect lots of different things. Temperature, distance, humidity, light, air quality, sound, hidden frequencies and more.
    • The AI part: Finally think about they kind of patterns you might see if you looked at all of that data in aggregate. Think about a simple way in which a computer might use that pattern to easily make a guess on what will happen next.
    • Also think about the various tones and lights and voices these devices use and how that makes you feel. Do you like the device? Do you trust it? Does it annoy you? Social robots are the same tech as a roaming laptop with eyes. But we see them much differently. Why? The personality of AI is going to be very important.
    • Now think of a product that lots of people would also need, invent it and become the first trillionaire..

OK, maybe it’s not quite that easy. But it’s not as hard as you might think.  No longer are the days where only big companies or trained scientists can invent the next breakthroughs. Every day people are solving problems and coming up with products from their garage or classroom.

Our future is not going to be exactly like The Jetsons – but still pretty amazing.  AI is here to stay and it’s just getting started. We don’t need to fear it we need to harness it.

Countless breakthroughs are coming and trillions in new revenue.

Hopefully because of someone just like you!

June 8, 2019 by Artificial Intelligence Basics of Robotics Machine Learning

Hot

Lynx Offroad Robot 

Unitree Robot Learning to Walk

Contact Us

Email support@robauto.ai