Banish the Darkness with Artificial Intelligence

/in , , , , , , /by

The Need

Most everything we do in life requires electricity. Transportation, farming, homes, businesses, computing—they all necessitate the efficient movement of electrons from a power production source to an energy use site¹. Unfortunately, since Thomas Alva Edison first illuminated a Menlo Park street in 1879 – thus banishing the darkness 2 – the  demands humanity has placed upon energy utilities have skyrocketed and consequently at least parts of our energy grid have become brittle and prone to failure. Moreover, the complexity of operating local, regional and national systems that are susceptible to cyber-attacks has made electricity a critical topic for national security.

World population has quintupled since the first sustained artificial light in the late 1800s. With more than 7.6 billion people 3 now teeming on our planet and all of us needing increasing amounts of energy 4, we cannot rely anymore on simple dynamo generators creating electricity and sending it straight to the desired location. On the contrary, our electricity grids are highly complex with numerous power sources (coal, gas, oil, nuclear, and renewables—biomass, solar, wind); switching and amplifying stations; transformers; above and below ground wiring; and storage batteries. Unsurprisingly, even considering conservative trends, our energy use is growing rapidly:

“In the New Policies Scenario, global energy needs rise more slowly than in the past but still expand by 30% between today and 2040. This is the equivalent of adding another China and India to today’s global demand.” 5

Moreover, our already highly electrified society will become even more dependent on charged electrons with new technologies. The Internet of Things (IoT), which requires cloud and edge computing at massive scale, is forecasted to exceed 75 billion devices connected to the internet by 2025 (up from 23 billion presently).6 Research into autonomous vehicles is rapidly approaching deployment stages. Testing underway in Phoenix, AZ and Pittsburgh, PA may be halted temporarily because of bad algorithms causing accidents or irregular policies, but other countries are not waiting around for the United States to perfect driverless cars. China is building cities specifically designed for autonomous vehicles.7 The Hyperloop – Elon Musk’s dream of accelerating people and cargo in a vacuum tube to near the speed of sound – is being tested in the United States and Europe.8

Population growth is also spurring construction of whole cities from scratch; examples abound in Asia. China plans on constructing a new city the size of Chicago (2.7 million people) with the latest technologies, and thus high energy demands.5 With support from Singapore, the Indian state Andhra Pradesh is building a new capital. “It will be a stupendous 7,235 square kilometres, 10 times the size of Singapore’s own 716 square kilometres.”9

The Risks

With so much riding on our need for energy, we must avoid costly failures. For example, when one author [TAC] worked in the semiconductor industry, the great Northeast Blackout of 2003 was tripped only a few miles from the production facility where he worked. The source of the power failure was overgrown tree branches that had fallen on a transformer. Normally, this would have remained a local issue, but the lead utility failed to react quickly enough to disengage the switches to other grids, thus causing a cascading, multi-grid failure for 50 million people from Ohio to New York to southeast Canada. People were trapped in subway cars and elevators for hours; cell phone service was disrupted for millions. 10 11

Perhaps even more dangerous than a lack of sensors on the grid and failed switches is the looming possibility of a cyber-attack targeting fragile points in the electrical system. Industry is spending billions in its attempts to strengthen cybersecurity of the brittle grid.12 “U.S. utilities will spend a cumulative $7.25 billion in security from now until 2020, with distribution automation assets as the core focus.”13 Grid security is a major issue within the US Government; simulations of cyber-attacks are routinely done by groups such as DOE and DHS.14 15

While a completely failure-proof grid may be impossible with the moving demand target caused by new technologies online and rapidly growing electricity demands, companies and government agencies must nevertheless do their best to anticipate and to prevent both accidental and nefarious situations before they happen. This presents a unique opportunity to engage newly accelerated capabilities of computers, especially artificial intelligence (AI).

The Opportunities

A core challenge with global electrical grids is their growing complexity. Growing exabytes of data from billions of IoT sensors, coupled with too-fast-for-humans reaction speeds required by increasing power demands, make a purely human-controlled electrical grid impossible. To sustain our growing needs, we must resort to a more digitalized approach.

There has been much talk recently about the so-called “smart grid,” defined as an electricity supply network that uses digital communications technology to detect and react to local changes in usage. 16 A smart grid has three major facets: data from sensors, computational power and optimized algorithms. Let us discuss the third point in detail.

Algorithm development has progressed in concert with Moore’s Law – the doubling of the number of transistors on a given semiconductor chip roughly every two years. In the early days of computers, there were a limited number of programming languages – Basic, Fortran, C/C++, Cobol, etc. Nowadays, companies such as Google develop their own proprietary operating systems as a routine course of smart business action. Increasingly, those algorithms are leveraging the power of AI.

While there are many definitions for AI, it can be defined simply as the ability of a machine to perform tasks commonly executed by a human. 17 AI is presently the hottest ICT 18 market sector. PwC estimates that global GDP will increase $15.7 trillion (a +14% boost compared to today) by 2030 as a result of AI. 19 20 Such a huge economic contribution cannot be understated, and organizations are investing aggressively accordingly. Venture capital funding pumped almost $5 billion into AI startups in 2017 alone. 21 AI has become a top corporate spending priority, with many hundreds of billions of dollars devoted to nabbing top talent and to securing algorithmic leadership. 22 There is a global race among governments to capture the title of AI world leader.  23

The reason for this flood of venture capital, corporate and government funding is that AI can solve problems far faster than humans, and in some cases solve problems that no human can. While we humans are smarter than any other living being on Earth, we are still rather dumb when it comes to absorbing and processing quickly vast amounts of data. The human mind is limited also to mostly working on a single task; despite some claims, we are generally horrible at multitasking. 24 Computers don’t have those limitations.

How might AI help the energy sector? We suggest there are two primary means by which advanced AI algorithms can improve efficiency, enhance safety, and improve the bottom line for energy utilities: grid optimization and cybersecurity.

Grid Optimization. To reap the benefits of the smart grid, AI will be a true necessity. “AI will be the brain of this future smart grid. The technology will continuously collect and synthesize overwhelming amounts of data from millions of smart sensors nationwide to make timely decisions on how to best allocate energy resources. Additionally, the advances made from ‘deep learning’ algorithms, a system where machines learn on their own from spotting patterns and anomalies in large data sets, will revolutionize both the demand and supply side of the energy economy.” 25

A challenge for utilities is maintaining consistent power. When there is a sudden increase in demand, the go-to reaction for utilities is to power-up ‘peaker plants’ that run on fossil fuels, generally natural gas. 26 Able to be turned-on within minutes of the detection of a demand spike – for example, due to increased air conditioning or heating requirements from a weather front moving through an area – peaker plants are unfortunately terribly inefficient and polluting. AI might assist in diminishing the use of peaker plants by leveraging advanced forecasting capabilities – for example, considering weather forecasts, regional demand cycles, and smart meter sensors in a single holistic package. The solution of such multidimensional problems is a strength of AI over humans. Such efforts could save utilities significant funds that they could funnel into further grid improvements.

One recent success story in using AI for energy optimization is that of Google’s deployment of algorithms to reduce electricity consumption in their server farms. Google has thousands of data centers worldwide to run search, store and process emails, etc. Those countless banks of GPUs 27 and memory storage systems consume tremendous amounts of electricity and tend to overheat. Because of their increasing presence and inherent inefficiencies, it is estimated that 2% of the world’s energy is consumed by data centers. 28 Both to save costs and to show-off its earlier purchase of the British AI startup DeepMind, Google decided to test AI in improving the cooling efficiency of their computer banks. “The autonomous AI control system initially led to a 12 percent improvement, which over nine months of operation increased to around a 30 percent improvement, with further improvements expected over time as its decisions are improved by having more training data. Google said in the long term that there is potential to apply the technology in other industrial settings.”29

Finally, AI can assist in converting incoming power to be acceptable to a grid. Every energy utility has a mix of power sources that it taps to provide its customer base their electricity. That mix is optimized based on price and, more recently, social perception for use of renewables. A challenge with multiple power sources – natural gas, oil, coal, nuclear, renewables – is to seamlessly integrate them all into the single grid. For example, wind and solar power sources require special converters to enable them to plug into the grid. “The high penetration of renewable energy systems is calling for new more stringent grid requirements. As a consequence, the grid converters should be able to exhibit advanced functions like dynamic control of active and reactive power, operation within a wide range of voltage and frequency, voltage ride-through capability, reactive current injection during faults, and grid services support.”30 AI could assist with energy conversion from renewables by leveraging deep learning to identify key performance optimization criteria and thus enable efficient uptake of the generated power into the smart grid.

Cybersecurity. Ironically, the very action of creating a smart grid makes it more vulnerable to cyber-attacks. Standardizing hardware and software, as well as creating a high degree of connectivity, enables hackers easier access to the grid through a greater variety of means. Programmed backdoors, firmware chip hacks, and even fake chips on motherboards 31 can all compromise the security of a given system.

Humans are inherently limited in what we can accomplish in cybersecurity. Our minds require rest, eating, drinking, etc.; cyber-attack bots require none of that. Thus, even though we might have the best intentions to be ever-diligent in the face of malware or phishing attacks, eventually we will slip up and allow nefarious actors access to sensitive computer systems.

Four years ago, seeing the rapid advances occurring in AI and knowing the challenges of human-centered cybersecurity, one author [TAC] forecasted that soon cybersecurity would enter the realm of “AI vs. AI,” i.e., AI algorithms would be both sources of attacks and defense. We are close to that time now. In the last few years, there has been a surge of interest in coupling the powers of AI, especially deep learning, to cybersecurity. 32 Thousands of companies globally now claim some form of AI in their cybersecurity offerings. Although it is debatable whether all these startups and corporate entities are truly using AI (some startups treat “AI” as a form of pixie dust—sprinkle it in a pitch in attempts to get funding), it makes sense to leverage the powers of learning algorithms to monitor and to react to cyber-attacks.

Ultimately, using AI for cybersecurity may not be just a choice but an imperative in the energy industry. In the United States, the complexity of the grid—with more than 8,000 power plants, 200,000 miles of high-voltage transmission lines, and 5.5 million miles of local distribution lines 33 34—necessitates that even local utilities must think hard about how they monitor their systems for intrusive cyber-attacks.

The Means

So how do utilities leverage the power (no pun intended) of AI? One challenge for cash-strapped utilities is that AI experts are in such demand that a starting salary for a star AI programmer can exceed seven dollar figures35. Thus, a more cost-effective approach may be partnerships among established startups and corporate entities that already have in-house expertise in AI. One such partnership was announced recently between BootstrapLabs (a leading venture capital firm focused on Applied AI in San Francisco) and innogy SE (a leading Germany energy company):

“BootstrapLabs and the innogy Innovation Hub will coordinate globally to build the largest artificial intelligence community for energy ecosystems, and provide a combination of capital and support to Applied AI startups that reimagine the future of energy production, distribution and management across decentralized and interconnected energy services for consumers, machines, enterprises, and public sector agencies.” 36

Sebastian Niestrath, SVP Infrastructure Platform Ventures at innogy New Ventures LLC, further clarified: “The need to interpret massive amounts of data and use AI-supported algorithms for grid operations is becoming increasingly important, especially in Germany – where electricity from renewable sources has pretty much quadrupled during the last 15 years and where millions of solar panels are now installed on residential buildings.” Moreover, with the California Energy Commission unanimously voting 5-0 in favor of mandating that all single-family homes, apartments and condominium complexes of three stories or less require solar panels as of January 2020, California will quickly follow suit in the need for AI-supported algorithms for grid operations.

Further such engagements among all major energy players – including utilities, AI experts, and local/regional/national governments – will be imperative as grid demands and threats increase.

The Future

As our population continues to grow and more technology gets plugged into existing grids, it will be incumbent upon utilities to increase efficiencies to remain cost-effective and to avoid grid failures. Moreover, the growing threat from cyber-attacks will demand that power companies better protect their grids from nefarious actors. AI is one approach that can significantly help utilities move into this challenging world. Opportunities will be strong and risks could be minimized for those leaders who embrace the capabilities of advanced computation.

Authors

Thomas A. Campbell, Special Advisor, BootstrapLabs and Founder of FutureGrasp, LLC

Thomas A. Campbell, Ph.D., is Founder and President of FutureGrasp, LLC (https://www.futuregrasp.com/), which advises organizations worldwide on trends and implications of emerging technologies.

Thomas is also serving as Special Advisor of BootstrapLabs, for Applied AI intellectual propriety development as well as assisting BootstrapLabs to expand its network reach and meet new opportunities into government sectors.

From February 2015 to August 2017, he was the first National Intelligence Officer for Technology (NIO-TECH) with the National Intelligence Council (NIC) in the Office of the Director of National Intelligence (ODNI). Tom’s insights have informed senior policymakers, enabled millions of dollars in industry and academic funding, broken ground in multiple new research areas, and kept diverse groups abreast of the rapid pace and implications of technology change.

Dr. Campbell is focused on emerging & disruptive science and technologies, especially identifying, tracking, forecasting and implications. He has extensive experience in government, academia, and industry – nationally and internationally. More about Tom at this link here.

Nicolai Wadstrom, Founder & CEO, BootstrapLabs

Nicolai is the Founder, CEO and Managing Partner of BootstrapLabs, a leading venture capital firm, based in Silicon Valley and focused on Applied Artificial Intelligence.

Nicolai has spent all of his professional life building technology companies. He started his first business at the age of 15 in the late 1980s and in 2008 he founded BootstrapLabs to build a scalable investment platform focused on being the most valuable partner for entrepreneurs to build successful companies.

With decades of operational and entrepreneurial experience, and having invested in, advised, and mentored over 30 companies, Nicolai works with investment decisions, and post investment mentorship to support scaling for BootstrapLabs’ portfolio companies.

BootstrapLabs is a Venture Capital company focused on Applied Artificial Intelligence that combines Venture Capital and Human Capital to build companies that define our future.

BootstrapLabs unique Venture Builder platform provides all the resources top-tier entrepreneurs need to take their companies from innovation to a Scalable Product Market Fit and from there to growth.

Prior to BootstrapLabs, Nicolai among other things founded a cutting edge Enterprise Software company (in the ECM and EAI space), he also co-founded and IPO’d the first CFD/FSB trading platform in the Nordics, and pioneered Virtual Reality in 1996. More about Nicolai at this link.

References:

  1. “Power is the capacity to use Energy…Power is like the strength of a weightlifter and Energy is the measure of how long he can sustain the output of power…Power is ‘watt’ and Energy is ‘watt-hour’.” http://www.differencebetween.net/science/difference-between-energy-and-power/#ixzz5Sze5vPtN 
  2. C. Klein, (December 17, 2014), “When Edison Turned Night into Day,” History, https://www.history.com/news/when-edison-turned-night-into-day
  3. “World Population by Year,” http://www.worldometers.info/world-population/world-population-by-year/
  4. The population considered in the middle class, and thus more prominent consumers of energy, is exploding globally. H. Kharas, February 17, 2017, “The unprecedented expansion of the global middle class,” Brookings, https://www.brookings.edu/research/the-unprecedented-expansion-of-the-global-middle-class-2/  
  5. “World Energy Outlook 2017,” https://www.iea.org/weo2017/
  6. “Internet of Things (IoT) connected devices installed base worldwide from 2015 to 2025 (in billions),” https://www.statista.com/statistics/471264/iot-number-of-connected-devices-worldwide/
  7. Kai-Fu Lee, “AI Super-Powers: China, Silicon Valley, and the New World Order,” 2018, Houghton Mifflin Harcourt, Boston.
  8. Hawkins, April 15, 2018, “World’s third hyperloop test track is now under construction,” The Verge, https://www.theverge.com/2018/4/15/17235262/hyperloop-transportation-technologies-test-track-france
  9. C. Cram, January 7, 2015, “Why Singapore is building a new Indian city 10 times its own size,” The Guardian, https://www.theguardian.com/public-leaders-network/2015/jan/07/singapore-building-india-city-andhra-pradesh
  10. History Editors, August 21, 2018, “2003 Blackout hits Northeast United States,” This Day in History, https://www.history.com/this-day-in-history/blackout-hits-northeast-united-states
  11. The plant where [TAC] worked had diesel generators that kicked-in immediately upon the power loss to avoid thousands of dollars per hour of manufacturing losses. The plant manager nevertheless called FirstEnergy Corporation quickly to find out what happened and when power would be restored. Their unbelievable initial response was, “Don’t worry, we’re on it. We’re watching CNN right now to figure out what happened.”
  12. C. Douris, September 21, 2017, “Utilities Will Spend Billions On Cybersecurity As Threat Grows,” Forbes, https://www.forbes.com/sites/constancedouris/2017/09/21/utilities-will-spend-billions-on-cybersecurity-as-threat-grows/#2dbe47b26cfe
  13. J. St. John, April 17, 2013, “Report: US Smart Grid Cybersecurity Spending to Reach $7.25B by 2020,” GTM, https://www.greentechmedia.com/articles/read/report-u-s-smart-grid-cybersecurity-spending-to-reach-7-25b-by-2020#gs.ZB7iBm8
  14. Department of Energy, Department of Homeland Security
  15. G. Bade, August 7, 2018, “Report: DOE, DHS planning new grid cybersecurity exercise this fall,” Utility Dive, https://www.utilitydive.com/news/report-doe-dhs-planning-new-grid-cybersecurity-exercise-this-fall/529518/
  16. https://www.google.com/ 
  17. Marr, February 14, 2018, “The Key Definitions Of Artificial Intelligence (AI) That Explain Its Importance,” Forbes, https://www.forbes.com/sites/bernardmarr/2018/02/14/the-key-definitions-of-artificial-intelligence-ai-that-explain-its-importance/#15b1f7be4f5d
  18. Information and Communication Technologies
  19. “Sizing the prize – PwC’s Global Artificial Intelligence Study: Exploiting the AI Revolution,” 2017, https://www.pwc.com/gx/en/issues/data-and-analytics/publications/artificial-intelligence-study.html
  20. PwC, Jue 27, 2017, “AI to drive GDP gains of $15.7 trillion with productivity, personalisation improvements,”  https://press.pwc.com/News-releases/ai-to-drive-gdp-gains-of–15.7-trillion-with-productivity–personalisation-improvements/s/3cc702e4-9cac-4a17-85b9-71769fba82a6
  21. J.D. Rowley, March 2, 2018, “Venture Funding Into AI And Machine Learning Levels Off As Tech Matures,” CrunchBase, https://news.crunchbase.com/news/venture-funding-ai-machine-learning-levels-off-tech-matures/
  22. Seitz, April 6, 2018, “Artificial Intelligence Becoming Top Corporate Spending Priority,” Investors Business Daily, https://www.investors.com/news/technology/artificial-intelligence-ai-spending/
  23. Minevich, December 5, 2017, “These Seven Countries Are In A Race To Rule The World With AI,” Forbes, https://www.forbes.com/sites/forbestechcouncil/2017/12/05/these-seven-countries-are-in-a-race-to-rule-the-world-with-ai/#69c6478b4c24
  24. N.K. Napier, May 12, 2014, “The Myth of Multitasking,” Psychology Today, https://www.psychologytoday.com/us/blog/creativity-without-borders/201405/the-myth-multitasking
  25. Wolfe, August 28, 2017, “How Artificial Intelligence Will Revolutionize the Energy Industry,” Harvard University, http://sitn.hms.harvard.edu/flash/2017/artificial-intelligence-will-revolutionize-energy-industry/
  26. “Peaking Power Plant,” https://en.wikipedia.org/wiki/Peaking_power_plant
  27. Graphics Processing Units, the current workhorse for deep learning AI.
  28. Pearce, April 3, 2018, “Energy Hogs: Can World’s Huge Data Centers Be Made More Efficient?,” Yale Environment 360, https://e360.yale.edu/features/energy-hogs-can-huge-data-centers-be-made-more-efficient
  29. Ranger, August 20, 2018, “ Google just put an AI in charge of keeping its data centers cool,” ZDNet, https://www.zdnet.com/article/google-just-put-an-ai-in-charge-of-keeping-its-data-centers-cool/
  30. Teodorescu, 2011, “Grid Converters for Photovoltaic and Wind Power Systems 1st Edition,” John Wiley & Sons, Ltd., https://www.amazon.com/Grid-Converters-Photovoltaic-Power-Systems/dp/0470057513
  31. Robertson, M. Riley, October 4, 2018, “The Big Hack: How China Used a Tiny Chip to Infiltrate U.S. Companies,” Bloomberg Businessweek, https://www.bloomberg.com/news/features/2018-10-04/the-big-hack-how-china-used-a-tiny-chip-to-infiltrate-america-s-top-companies?srnd=premium&mod=djemceocouncil
  32. “Applied AI Conference 2018 – Panel – Applied AI and Cybersecurity – Making the Enterprise More Secure,” BootstrapLabs, https://www.youtube.com/watch?v=4ckuyHpyA00
  33. December 8, 2017, “How many power plants are there in the United States?,” https://www.eia.gov/tools/faqs/faq.php?id=65&t=2
  34. J. Weeks, April 28, 2010, “U.S. Electrical Grid Undergoes Massive Transition to Connect to Renewables,” Scientific American, https://www.scientificamerican.com/article/what-is-the-smart-grid/
  35. C. Metz, April 19, 2018, “A.I. Researchers Are Making More Than $1 Million, Even at a Nonprofit,” New York Times, https://www.nytimes.com/2018/04/19/technology/artificial-intelligence-salaries-openai.html
  36. Press Release, October 3, 2018, “innogy Innovation Hub and BootstrapLabs to Form a Partnership for Investments in AI and Energy,” San Francisco Business Times, https://www.bizjournals.com/sanfrancisco/prnewswire/press_releases/California/2018/10/03/NY26733