A Peek into AI and what it means to us
AI is the new electricity. Just as 100 years ago electricity transformed industry after industry, AI will now do the same.
The above insight by the deep learning pioneer Andrew Ng in 2016 has clearly shown us a path that AI will take in the next few decades. As a junior software engineer, I am curious about where AI will lead humans to and more importantly, what changes and how the changes will be taken place?
In the book “AI Superpowers” by one of the brightest minds in the realm of AI, Kai-Fu Lee, he points out that we are coming to the age of implementation where no breakthrough in the algorithm is needed but a group of good-enough engineers will be sufficient to revolutionize industries.
This concept resonates with the idea of “Zero to One” by Peter Thiel, the billionaire entrepreneur. The essence is that it is always extremely hard to create something out of thin air, like creating a search engine when there’s no search engine on the Internet. But once there’s a pioneer in the field succeeded, the followers could relatively easy duplicate the process to achieve the same result.
In the book, Kai-Fu Lee indicates the global shift from age of discovery to the age of implementation. The world of AI has gained tonnes of progress from a group of elite AI scientists while now the industries are ready to get their hands dirty to play the algorithms released by elites to match the specific needs in different fields to maximize the financial reward.
To implement the AI algorithms successfully, he thinks there are three essential components: big data, computing power, and the work of strong AI algorithm engineers. When the computing power and the talent level reach a certain acceptable threshold, Kai-Fu Lee believes the quantity and quality of data will play a decisive role when comes to the accuracy of AI algorithms.
While the United States is taking the lead in 2018 (the year the book was published), he predicted China will leapfrog in 4 different areas of AI in the light of government support, China entrepreneurs a.k.a. “gladiators in the Coliseum” and the enormous amount of data provided by the people of China.
These 4 areas of AI have their respective implementations and business cases to explore.
This area refers to AI algorithms developed by tech giants like Netflix, Google or Baidu. These behemoths can precisely predict the movie we like, the best search result for us contributed from their internal smart recommendation system.
When Internet AI largely benefits the tech giants, the business AI allows SMEs to non-tech MNCs to exploit the data they sit on. Before the dawn of AI, traditional businesses had a hard time finding insightful patterns from their data lake to further increase revenue. Not in this era. Different sectors like insurance and healthcare can empower their business strategies with the help of Business AI. For example, Business AI could help exploit data to generate insights for companies.
And then here comes another facet of AI: Perception AI. It allows machines to interpret images and audio by correlating meaning to identified objects in images or sentences in a recording. By having the ability to perceive the media, meaningful data could be collected to make a better judgement or to apply business logic.
Perception AI essentially improve the ability of machines to sense the surrounding environment. Now AI can make rational movements with robotic arms and any other extensions. Unlike the first three areas, this fourth area requires the combination of intelligence from hardware to make the most out of Autonomous AI. Drone, self-driving cars and autonomous machines in factories are the potential products of this area.
Artificial General Intelligence (AGI)
Kai-Fu Lee told us in the book that the 4 areas of AI mentioned above are reachable to our extent. But, above these 4 areas, what will happen if machines can perform every task that humans can do, or even outperform humans?
That is the overall idea of Artificial General Intelligence which is far from our reach. The idea splits the crowd in the AI community into Utopian and Dystopian.
The utopians believe that AGI would be able to help humans conquer the most difficult questions of humanity on the universe, incurable diseases and more. It would make humans leapfrog to the Type 1 civilisation of Kardashev scale in the blink of an eye.
On the other hand, we have people like Elon Musk and Stephen Hawking warned us on the development of AI.
Unlike the movie Terminator, AI may not wipe humanity with robots. What could possibly happen is humans make an instruction that destroys civilisation by accident. Imagine a superintelligent being instructed to solve global warming no matter what is the cost. The scientists fail to set the constraints right on what to not do to achieve the goal. The superintelligent could shut down the power plants or spread diseases to ensure there are no humans on earth: because it thinks the human is the root cause of global warming!
This issue has a name: value alignment problem.
What to do
At the moment, I believe AI do more good than harm. Its algorithms have helped humans to achieve more and we all can enjoy a better quality of life.
Ever heard of Copilot? As a junior software engineer, an AI pair programmer could be my saviour to keep up with the speed to ship features on time. I think I am going to try it out soon.
With my limited exposure to AI, I am still unclear about what AI could bring to the table. But we could keep updated on news and actively involve ourselves in this field.
If you are interested to know more about how we could do to safeguard the development of AI, I recommend this article “Positively shaping the development of artificial intelligence” by 80,000 hours organization.
Have a great day.