AI. What it will and wont do & the challenges ahead

Epic just unveiled Unreal Engine 5 and it looks amazing! Intel Labs is working with Grand Theft Auto on a new machine learning project called Enhancing Photorealism Enhancement and the footage looks unreal. It will be hard to distinguish video games from real life soon, which got me thinking….Where are we with AI? We’ve heard a lot of promises…….

What is AI doing and what is it not doing and what lies ahead?

Right now we have narrow AI. This is when systems perform very specific tasks. Think of this as the filtering of your email for junk/spam, or your favorite map app guiding your way or interfacing with Alexa/Siri.

We are getting broader with such advancements in chatbots, logistics, self-driving cars, virtual doctors, virtual nursing assistants, tutors, music making, art producing and more. The holy grail is still general AI. This is where AI has the ability to understand or learn any intellectual task that a human being can.

Don’t freak out…we are not even close….

Let’s look at what AI is doing

You’ve probably read the headlines of Microsoft’s recent acquisition of Nuance. A big area for AI is natural language processing. We have a long way to go but when algorithms finally crack natural language, many business will have substantial use cases.

• Chatbots
• Market Analysis
• Instant Translations
• Resume Readers
• Phone Auto-Attendants
• Drive-thru order taking
• Video Games

Part of the advancement has come from GPT-3. One of the coolest uses of this has been with the real estate market. AI is now writing listings for properties across the globe.

• Entertainment has been using AI for years. Your favorite streaming platform uses data and algorithms to offer you suggestions on what to watch next.
• Farming has adopted AI to reduce chemical inputs. Machinery is able to distinguish between weeds and non-weeds and then chemicals are only applied to weeds.
• I heard recently on a coding podcast where someone suggested that we should have Grammarly for code :) Code reviews can be automated. Facebook is even using AI to convert code from one language to another. How is it that Grammarly is better at language than we are?
• AI is being used by video game developers to identify cheaters. What they should do is develop AI to determine skill levels of players and match them accordingly. It should be easy to distinguish movement, accuracy, game sense, reaction time, etc… to organize players by skill and ready them up in appropriate lobbies. Separate the smurfs, trolls, casual players, and competitor players so everyone can have an equal, fair and enjoyable player experience.
• Healthcare has embraced AI for diagnosing chronic conditions, aiding in complicated surgical procedures, treatments all using medical data.
• AI continues to help with the Covid crisis. It has assisted with early diagnosis, forecasting and tracking cases as well as tracing contacts. It was responsible for the various treatments as well in record time. This has paved the way for speeding up the process of drug discovery going forward. Nurse bots are being used to monitor patients to reduce the burden when there are more patients in the hospital.
• Finally education has embraced AI for teaching aids such as grading papers which reduces the amount of workload for teachers. It has been able to personalize learning based on strength and weaknesses of each student.

What AI is not doing

• Self-driving cars. We are decades away.
• Moderation on forums and social media are still armies of human beings scrubbing content.
• Bank tellers. Let’s look at ATMs. ATMs have been around for decades. They could have easily replaced human tellers. There are more tellers now than when ATMs were widely released. How? ATMs lowered the cost of opening bank branches, so they opened more which required hiring more tellers. Plus, let’s be honest, the ATM doesn’t help with everything you need to do at your bank especially as they offer more and more services
• Language translation. We will nail this very soon. But why are human translators demand skyrocketing? If the cost and ease of basic translation drops to nearly zero, the cost of doing business with those who speak other languages falls. More companies will do more business overseas, creating more work for human translators. AI may do the simple translations, but humans will close the deal!
• Replace massive jobs. It is easier to see existing jobs disrupted by new technology than to envision what new jobs the technology will enable. Technology has progressed nonstop for 250 years, and in the US unemployment has stayed between 5 to 10 percent for almost all that time, even when new technologies like steam power, electricity and the Intern emerged. With the Internet we got thousands of new companies worth trillions of dollars. Dozens or new careers emerged. The cost of starting a new business plummeted, and the cost of communicating with customers and leads went to nearly zero.

Yes, we mail fewer letters, but we have a lot of companies employing a lot of people to now send emails and messages to one another. We buy fewer newspapers, but the digital pivot employs thousands of reporters and bloggers online.

You will just do different things because much of what you do today may be automated. During the industrial revolution farmers turned into factory workers and then factory workers became factory managers, and so on…This is how free economies work.

AI allows everyone to be even more exceptional with new tools in their hand.

What challenges lie ahead?

Companies are asking AI to solve the wrong problem. They want AI to make decisions for them. This will only be useful for decision making where more information rarely changes the organizational politics behind a decision.

AI will never replace humans. An example is education. We all want autonomous tutors. the AI tutor will monitor student progress and understand what motivates them and provide an adaptive learning experience. Current algorithms can’t read motivation and are from it. We must be mindful of reasons outside of just knowledge acquisition. We want to develop not only our brain but our emotional IQ and skills, human mentor-ship and human connections in our communities. We all remember a teacher or two who changed our lives. Will we remember what Siri said?

Loneliness is on the rise. One study found a relationship between depression among adolescents and screen time, compared to youth who spent time on off-screen activities with in person activities. We have to be conscious of driving socioemotional goals of education and not compel children towards screens.

We also have to be careful with the misuse of AI to extend power. We don’t want to teach violent subject matter. We don’t want social media platforms to amplify destructive and democratic organizing. What will governments and dishonest companies do with the personal data processed from these AI platforms?

Scaling a centralized model reduces the number of voices who decide what is taught. The choice of who gets to teach becomes a choice of who gets to learn. We need to ensure that it does not become homogeneous.

Finally, we have to be careful of data privacy and cyber security. There was a recent example of an online mental health therapy company who was compromised and very personal data patient data was leaked online for anyone to consume.

AI is very good at focusing on automation of tasks and development of exciting problems, rather than replacing humans. Use AI to understand process. AI monitors the process, while the teacher and manager remains the coach. A hybrid model elevating mentors and coaches.

AI supports humans, not replace them.

Many professions will experience faster than average growth not in spite of AI, but through it. Just as we saw with the internet, real gains in jobs will come from places where our imagination and creativity cannot yet take us.