Thumb1

In 2019, I wrote about the top 10 people you need to follow if you care about AI[1].  Recently, two institutions mentioned in my previous article made promising breakthroughs that are worth looking into today because they may lead to noticeable technological disruptions.

Technology disruptions are not historical anomalies, but rather consistent events with an unpredictable frequency. Some examples of major disruptions are; utilizing steam power to fuel the industrial revolution in the 19th century, wide adoption of electricity in the beginning of the 20th century, then in the 21st century humans revolutionized information with the internet.

Two commercial institutions working on AI made promising breakthroughs

In the past decade, there has been an overwhelming abundance of available information to humans because of the internet, coupled with extraordinary human research activity, and constant increase in productivity. This led to the development of many new technologies that make it hard to predict which ones are going to be disruptive and which ones would be the dominant technologies of the future. Trends like cellphones, social media, cloud computing emerged and newer technologies like wearables, streaming, augmented reality, blockchain are starting to make their way into the mainstream.

‘’Technology disruptions are not anomaly events, but rather consistent with an unpredictable frequency’’

However, for the common non-techy crowd, there are far more important technology disruptions that are fundamentally changing and will continue to change the future. Renewable energy development is disrupting the energy world; with falling capital costs for solar and wind energy[2] coupled with battery energy storage[3]. The end result may be the lowest costs of energy humans have ever had, the disruptive result may become that energy, like information, may be provided to the end consumer at little to no cost. Another fundamental disruption is advent of near zero marginal cost for production[4] of high volume consumables, this translates to very low cost items to the end user, and may eventually result in little to no cost to consumers for such items. This has a disruptive effect on how people live.

When it comes to the overhyped topic of Artificial Intelligence(AI)[5] the non-techy crowd could care less about the technicals, however, most people will be living through fundamental disruptions to their lives that are powered by AI technology disruptions. So far AI-based computer code is wildly successful in affecting the daylife of may people in the following fields: translation (google translate[6]), recognition (speech recognition: siri[7] call bob, alexa[8] buy toilet paper), and classification (what will appear on your facebook/instagram/news feed). However, a new era of disruption is coming with OpenAI’s Generative Pretrained Transformers (GPT) and Google’s Deepmind Alpha series of programs (AlphaGO, AlphaZero, AlphaStar).

“Open AI allowed the use of GPT-3 by select limited number of people”

First let’s start with Open AI, a company whose mission is to ensure that artificial general intelligence (AGI) i.e. highly autonomous systems that outperform humans at most economically valuable work, benefits all of humanity. The company gave access to the use of a computer program (via an API) called GPT-3 to a select limited number of people, and announced it will share it with Microsoft. GPT-3 is a type of a computer program called a transformer[9] that is trained on a language data set from the internet. It has the ability to transcribe language based on a prompt given to it. The output transcription has a meaningful relationship to the input text. GPT-3 can provide answers to questions, comments on phrases, and perform a writing task written to it. The implications are substantial because if used by anyone it could flood the internet cyberspace with text that has meaning, but created by the program. Essentially introducing more noise to human to human communication within the internet, internet noise is not a new problem though, in the recent years language bots became popular over the internet, however, they were easy to recognize by humans and to detect by automated systems, because these primal bots didn’t provide much of meaning or material to the reader. As a result, they can be easily reported and discarded by humans, and detected and controlled by anti-spam systems. The challenge with GPT-3 is that In a zero-day scenario[10], GPT-3 texts are not easily recognized by humans and in some cases not recognized at all. This is a very problematic issue in terms of risks to meaningful communication on the internet in addition to posing risk to the public decorum in cyberspace. In a nutshell, OpenAI created GPT-3, a program that can be very helpful depending on how it is used, but didn’t make it “open” to everyone. Eventually GPT-3 will be available in the wild and the opportunities for its use will be limitless. This technology will be disruptive. This promising technology will help in facilitating human communications, augmenting innovation, and elevating arts.

Open AI announced they will share GPT-3 with Microsoft

Now moving to Google, Deepmind (a Google company) has been doing interesting work trying to build a program that can play games, any game, they started with AlphaGo, which demonstrated superiority to humans in the compex ancient chinese game of Go, then they moved to creating AlphaZero, the program played chess and became a super chess player. Later, the Deepmind team did the right thing by expanding into another domain to solve a real world problem. They developed AlphaFold, a program that predicts protein folding with high accuracy, protein folding is a complex biological phenomenon. Accurate protein folding[11] prediction is a hard physical/biological problem to solve, and the Deepmind team just demonstrated the capability of solving this problem using AlphaFold 2. This will open the door for a whole new era of advancement in biotechnology. This will be disruptive technology when it is allowed to be used by others. This promising disruption may lead to big advancements in medicine, improving public health, extending human life expectancy, and enhancing human abilities.

‘’There not enough evidence that shows that Google’s Deepmind’s technology can be transferred seamlessly between domains’’

Protein Folding Explained

On a final note, it is interesting to see the differences between Open AI and DeepMind approaches in using AI in real world applications. Deepmind has an approach that seems to be more focused to certain domains (games and recently Biology), however, there is not enough evidence that shows that the technology can be transferred seamlessly between domains, which is an issue for Deepmind and a relief for anyone who is concerned about AI. On the other hand, while OpenAI’s GPT-3 has its limitations, it is focused on solving language which is a universal knowledge domain, so it can be used across domains, this seems to be in line with the philosophy of: solving language solves everything else. It will be interesting to see what a GPT-4 , GPT-5 or eventually GPT-10 can do. This strategy may fail or take longer time to succeed. But in the meantime the DeepMind team and humans will reap the benefits of the Alpha’s algorithms in solving practical problems.

Reference

[1] “10 People you should follow if you care about AI | by Mike ….” https://medium.com/@7asabala/10-people-you-should-follow-if-you-care-about-ai-40d94081697d. Accessed 26 Dec. 2020.

[2] “How Falling Costs Make Renewables a Cost-effective ….” https://www.irena.org/newsroom/articles/2020/Jun/How-Falling-Costs-Make-Renewables-a-Cost-effective-Investment. Accessed 26 Dec. 2020.

[3] “Understanding Battery Energy Storage Systems and Their ….” https://www.nrel.gov/usaid-partnership/project-understanding-battery-storage-grid-integration.html. Accessed 26 Dec. 2020.

[4] “Toward a zero marginal cost society | The Japan Times.” 24 Jan. 2018, https://www.japantimes.co.jp/opinion/2018/01/24/commentary/world-commentary/toward-zero-marginal-cost-society/. Accessed 26 Dec. 2020.

[5] More precisely machine learning based computing algorithms :  “Artificial Intelligence (AI) Definition – Investopedia.” 13 Mar. 2020, https://www.investopedia.com/terms/a/artificial-intelligence-ai.asp. Accessed 26 Dec. 2020.

[6] “Google Translate.” https://translate.google.com/. Accessed 26 Dec. 2020.

[7] “Siri – Apple.” https://www.apple.com/siri/. Accessed 26 Dec. 2020.

[8] “Amazon Alexa Official Site: What Is Alexa?.” https://developer.amazon.com/alexa. Accessed 26 Dec. 2020.

[9] “How Transformers Work. Transformers are a type of neural ….” 10 Mar. 2019, https://towardsdatascience.com/transformers-141e32e69591. Accessed 26 Dec. 2020.

[10] “Zero-day vulnerability: What it is, and how it works – Norton.” 28 Aug. 2019, https://us.norton.com/internetsecurity-emerging-threats-how-do-zero-day-vulnerabilities-work-30sectech.html. Accessed 26 Dec. 2020.

[11] “AlphaFold: a solution to a 50-year-old grand challenge in ….” 30 Nov. 2020, https://deepmind.com/blog/article/alphafold-a-solution-to-a-50-year-old-grand-challenge-in-biology. Accessed 26 Dec. 2020.

Previous articleFiltering Out Ideas Made Me More Productive
Next articleWhy Investors Should Consider Buy and Hold Real Estate
Mike Hassaballa
Mike earned a master’s degree in applied science in 2013, then he launched his career in the data centre industry. In 2015, he shifted gears and took on a Lead Engineer role in a company developing emission reductions technology. He then moved in 2018 into energy consulting. Mike focuses on most critical issues and opportunities in business: strategy, operations, technology, transformation, advanced analytics, and sustainability. Mike writes fascinating stories meant to be read by anyone. He excels in simplifying complex subjects and bringing a fresh new perspective to pressing issues.

LEAVE A REPLY

Please enter your comment!
Please enter your name here