Seriously, are you not bored of people like Elon Musk or Stephen Hawking, who are frequently claiming that artificial intelligence (AI) is a big threat to human society? Those discussions let people think AI is going to steal everyone’s job and start the next world war. Even worse, it fuels expectations to a point that the general public is getting the idea that AI research has recently made quantum leaps. And thanks to raising these expectations we could easily run into an AI winter again.
Keep calm: we are far away from creating a super-intelligence Musk and Hawking are talking about or what Hollywood is showing us with movies like Ex Machina, Her or AI. Let me tell you something that should concern you more. According to a Sage survey, 43 percent of respondents in the U.S. and 46 percent of respondents in the UK answered that they have no idea what AI is about. Thus, the biggest issue these days is, that the general public has no clue what AI really means.
It is absolutely important to talk about regulations and ethics in terms of AI. Indeed, Boston Dynamics’ robots “Atlas” or “Handle” are spooky. But what if Tesla’s autopilot is knocking somebody down – who is responsible? Or if the autopilot has to decide which person should survive. The driver? The old man or the young boy on the street? These are the main questions I am getting frequently from the audience. This is what people are concerned about.
Right now, we are living in a time of exponential technological developments. Meaning, “this will never happen” is the wrong attitude since technology research and progress is happening in ever so short timeframes. However, the most successful AI systems today are solving the game of Go or the strategy game FreeCiv.
Today, most AI projects are based on machine learning that simply helps to identify patterns within data sets and thus tries to make predictions based on the existing data. However, having the right data and quality data is what’s most important. And then afterwards to check the plausibility and correctness of the results since you can always find something in endless sets of data. And that’s also one of the drawbacks if you consider machine learning as a single concept. Machine learning needs lots of training data to learn and be able to find valuable information respectively results in patterns.
Artificial intelligence is autonomous process automation
To bring the AI discussion to reasonable deliberations. Today, AI is not about recreating the human brain. It is about building a system that acts like a human. All in all, AI means analytics, problem solving and autonomous automation based on data, knowledge and experience.
Just try to look at this entire AI discussion from a different angle. Consider our life as a process. Consider every day as a process that is divided into single steps. And then, consider AI as autonomous process automation that provides us with more convenience and thus makes our life easier.
Amazon Alexa and Apple Siri are AI but not intelligent
Have you ever tried to lead a simple conversation with Amazon Alexa or Apple Siri? Exactly, this is not going to go well. However, Alexa as well as Siri are AI technologies. They use natural language processing (NLP). So, machine learning algorithms powered by predictive models. For example, the algorithms are used to translate your commands into little pieces, so-called sound bites. Afterwards, these pieces are analyzed by another predictive model, that tries to predict what your request was about. However, Alexa or Siri are not intelligent or self-learning. Consider their “brains” (they do NOT have brains) like a “database” in the cloud-backend of Amazon and Apple operating a set of ready answers or action items. If you are a proud owner of an Amazon Echo device you will understand what I mean. Every Friday you get an email with the newest commands you can use to control Alexa. Besides the ever-growing database behind Alexa, the so-called “Alexa-Skills” are helping Alexa to become more “intelligent.” This is nothing else but small applications (like for your Android smartphone or iPhone) someone has to develop and equip with further commands, questions you can ask, ready answers or action items. And the more skills are activated, the more intelligent Alexa appears since you have more commands and ways to interact with her. However, the interesting story around Alexa is that Amazon has 5,000 people exclusively working on Alexa to enhance it. So, we could expect more progress soon.
The good news, self-learning systems already do exist. If you connect your iPhone via Bluetooth with your car and if you’ve set up your work and home address, soon the iPhone will tell you how long you are going to need to work or back home. In some other cases “Apple Maps Destination” has shown me predictions to locations even if I didn’t store them exclusively on my iPhone. So, just based on my travel habits. Google Now works the same way. It proactively provides information to the users they are maybe looking for. Predictions based on their search habits. However, if you are granting Google access to your calendar, Google Now may also work as an advisor. It reminds you that you have an appointment and what vehicle you should take in order to arrive at your destination on time.
Other AI-related services you might have been in touch with for quite some time are:
- Since the 1950s, the financial service industry is using machine learning algorithms for credit scoring.
- Dating websites use algorithms to provide an indication of how likely two people are a match and would probably go out on a date.
- At Geneva Airport, autonomous robot KATE takes care of your luggage, checks you in and guides you through the airport. Everything based on data analytics and geo location analysis.
- “Relay” from Savioke is an autonomous delivery robot that is used in some hotels for room service purposes.
Autonomous process automation in your daily life
Back to AI as an autonomous process automation that makes our life easier. Let’s discuss some ideas:
- Imagine Alexa or Siri as your personal watchdog in business. Checking your phone calls, autonomously negotiating appointments with e.g. colleagues – especially with the ones who always send calendar invitations before asking. I am kind of talking about a very early version of Iron Man’s AI “Jarvis.”
- Or what about Alexa or Siri as your very personal life assistant. Let’s say you are speaking at a conference. Your flight home leaves at 4 p.m. In order to reach your flight on time, your personal life assistant autonomously orders you a ride for 2:15 p.m. since your speech ends at 1:45 p.m. and the traffic information looks busy these days. The virtual assistant just sends you the details of the pickup location which is right in front of the conference venue. The assistant simply follows the entire process you would usually go through: from taking the phone out of your pocket, opening the app, searching for the target location and then ordering the ride. So, keep your hands and mind free for more important things. Of course, you have to give the AI access to your calendar, geolocation and other information.
- Or imagine an intelligent version of the kitchen aid “Thermomix.” The dish advisor: Based on what the kitchen helper finds in your fridge it makes recommendations what you could possibly cook. And when some ingredients for a dish are missing it could ask for permission to order them online. The health advisor: Based on your eating behavior of the last weeks it would nicely recommend you not to cook the tiramisu you were just about to prepare, because this wouldn’t be good for your consumption of calories.
Autonomous process automation inside your organization
AI as an autonomous process automation has lots of potentials inside an enterprise as well. Let’s focus on two real world examples:
- An AI-defined infrastructure is actually nothing else but autonomous process automation for IT operations. Those kinds of environments are capable of deploying the necessary resources depending on the workload requirements as well as de-allocating the resources when they are not needed anymore. They are constantly analyzing the ever-changing behavior and status of every single infrastructure component and thus understanding the environment they are acting in. The environment is reacting or proactively acting based on the status of single infrastructure components by autonomously taking actions and thus leading the entire infrastructure into an error-free status.
- As part of business operations, autonomous process automation is already utilized in the insurance industry, autonomously creating insurance policies. In doing so, experts are teaching the AI the process, step by step, how and why they create a certain insurance policy. Besides that, the AI is also connected to partner systems that deliver further information. For the time being, the AI creates an insurance policy and hands it over to an expert. The expert checks the insurance and tweaks it if needed. After that it is delivered to the customer. In the end, a customer will directly interact with the AI and, after delivering the necessary information, will receive his/her custom insurance policy.
Keep your expectations low
On a final note. Instead of constantly discussing AI as the biggest threat to humanity, maybe we should talk about the biggest threat to AI research. The impatience of humanity! The issue: AI research has been an oscillating system between several techniques. Whenever one approach did not do “the job completely,” people get frustrated and turn to another one. You want to see amazing AI stuff in the future? Keep calm and wait. It will happen! We just have to move forward step by step. Otherwise our expectations will fail and will immediately guide us into the next AI winter.
Saying this, consider AI technologies as an approach to make your life easier. And don’t stress the term “intelligent” too much. However, always be aware of what type of data and personal information you really want to share. Once it is out, it’s virtually impossible to roll it back.