White House Chronicle

News Analysis With a Sense of Humor

The Backdoor Challenge of AI Machine-Learning

by Leave a Comment

The great race is on. It isn’t the one on television, but it is one that has put the world’s wealthiest companies in fierce competition to secure market share in artificial intelligence.

The handful of big-tech companies and their satellites may have spent as much as $1 trillion on machine-learning and data center infrastructure to stuff their AI systems with billions of bits of information hoovered up from public and private sources on the internet.

These companies — Amazon, Google, Meta, Microsoft and OpenAI among them — are rich and have made their creators rich beyond compare because of information technology. Their challenge is to hold onto what they have now and to secure their futures in the next great opportunity: AI.

An unfortunate result of the wild dash to secure the franchise is that the big-tech companies — and I have confirmed this with some senior employees — have rushed new products to market before they are ready.

The racers figure that the embarrassment of so-called hallucinations (errors) is better than letting a competitor get out in front.

The challenge is that if one of the companies — and Google is often mentioned — isn’t on the leaderboard, it could fail. It could happen: Remember “MySpace”?

The downside of this speedy race is that safety systems aren’t in place or effective — a danger that could spell operational catastrophe, particularly regarding so-called backdoors.

According to two savants in the AI world, Derek Reveron and John Savage, there is a clear-and-present danger presented by this urgency for market speed over dangerous consequences.

Savage is the An Wang professor emeritus of computer science at Brown, and Reveron is chair of the National Security Affairs Department at the Naval War College in Newport, Rhode Island.

Reveron and Savage have been sounding the alarm on backdoors, first in their book, “Security in the Cyber Age: An Introduction to Policy and Technology,” published by Cambridge University Press early this year, and later in an article in Binding Hook, a British website with a focus on cybersecurity and AI.

“AI systems are trained neural networks, not computer programs. A neural net has many artificial neurons with parameters on neuron inputs that are adjusted (trained) to achieve a close match between the actual and the desired outputs. The inputs (stimuli) and desired output responses constitute a training set, and the process of training a neural net is called machine-learning,” the co-authors write.

Backdoors were initially developed by telephone companies to assist the government in criminal or national security cases. That was before AI.

Savage told me that backdoors pose a grave threat because, through them, bad actors can insert malign information — commands or instructions — into computers in general and backdoors in machine-learning-based AI systems in particular. Some backdoors can be undetectable and capable of inflicting great damage.

Savage said he is especially worried about the military using AI prematurely and making the nation more vulnerable rather than safer.

He said an example would be a weapon fired from a drone fighter jet flying under AI guidance alongside a piloted fighter jet where the weapon fired by a drone could be directed to do a U-turn and come right back and destroy the piloted plane. Extrapolate that to the battlefield or to an aerial bombardment.

Savage says that researchers have recently shown that undetectable backdoors can be inserted into AI systems during the training process, which is a new, extremely serious, and largely unappreciated cybersecurity hazard.

The risk is exacerbated because feeding billions of words into big-tech companies’ machine-learning systems is now done in low-wage countries. This was highlighted in a recent “60 Minutes” episode about workers in Kenya earning  $2 an hour, feeding data to machine-learning systems for American tech companies.

The bad actors can attack American AI by inserting dangerous misinformation in Kenya or in any other low-wage country. Of course, they can launch backdoor attacks here, too, where AI is used to write code, and then control for that code is lost.

In their Binding Hook article, Reveron and Savage make a critical point about AI. It isn’t just another more advanced computer system. It is fundamentally different and less manageable by its human masters. It lacks an underlying theory to explain its anomalous behavior, which is why the AI specialists who train machine-learning systems cannot explain this behavior.

Deploying technology with serious deficits is always risky until a way to compensate for them has been discovered. Trouble in is trouble out.

What Might Happen If Google Is Broken Up

by Leave a Comment

Alphabet Inc.’s Google has few peers in the world of success. Founded on Sept. 4, 1998, it has a market capitalization of $1.98 trillion today.

It is global, envied, admired and relied upon as the premier search engine. It is also hated. According to Google (yes, I googled it), it has 92 percent of the search business. Its name has entered English as a noun (google) and a verb (to google).

It has also swallowed so much of world’s advertising that it has been one of the chief instruments in the humbling and partial destruction of advertising-supported media, from local newspapers to the great names of publishing and television. All of these are suffering and many have failed, especially local radio and newspapers.

Google was the brainchild of two Stanford graduate students, Larry Page and Sergey Brin. In its short history, it has changed the world.

When it arrived, it began to sweep away existing search engines simply because it was better, more flexible, amazingly easy to use, and it could produce an answer from a few words of inquiry.

Seven major search engines were fighting for market share back then: Yahoo, Alta Vista, Excite, Lycos, WebCrawler, Ask Jeeves and Netscape. A dozen others were in the market.

Since its initial success, Google — like Amazon, its giant tech compatriot — has grown beyond all imagination.

Google has continued its expansion by relentlessly buying other tech companies. According to its search engine, Google has bought 256 smaller high-tech companies.

The question is: Is this a good thing? Is Google’s strategy to find talent and great, new businesses, or to squelch potential rivals?

My guess is some of each. It has acquired a lot of talent through acquisition, but a lot of promising companies and their nascent products and services may never reach their potential under Google. They will be lost in the corporate weeds.

During its acquisition binge, Google has changed the nature of tech startups. When Google itself launched, it was a time when startup companies made people rich when they went public once they proved their mettle in the market. Now, there is a new financing dynamic for tech startups: Venture capitalists ask if Google will buy the startup. The public doesn’t get a chance for a killing. Innovators have become farm teams for the biggies.

Europe has been seething about Google for a long time, and there are moves to break up Google there. Here, things were quiescent until the Department of Justice and a bipartisan group of attorneys general brought a suit against the company for monopolizing the advertising market. If the U.S. efforts to bring Microsoft to heel are any guide, the case will drag on for years and finally die.

History doesn’t offer much guidance about what would happen if Google were broken up. The best example and biggest since the Standard Oil breakup in 1911 is AT&T in 1992. In both cases, the constituent parts grew faster than the parent. The AT&T breakup fostered the Baby Bells — some, like Verizon, have grown enormously. Standard Oil was the same: The parts were bigger than the sum had been.

When companies have merged with the government’s approval, the results have seldom been the corporate nirvana prophesied by those urging the merger, usually bankers and lawyers.

Case in point: the 1997 merger of McDonnell Douglas and Boeing. Overnight, the nation went from having two large airframe manufacturers to having just one, Boeing. The price of that is now in the headlines as Boeing, without domestic competition, has fallen into the slothful ways of a monopoly.

Antitrust action against Google has a few lessons to be learned from the past. Computer-related technology is just too dynamic; it moves too fast for the past to illustrate the future. That would have been true at any time in the past 20 years (the years of Google’s ascent), but it is more so now with the arrival of artificial intelligence.

If the Justice Department succeeds and Google breaks up after many years of litigation and possible legislation, it may be unrecognizable as the Google of today.

It is reasonable to speculate that Google, at the time of a breakup, may be many times its current size. Artificial intelligence is expected to bring a new surge of growth to the big tech companies, which may change search engines altogether.

Am I assuming that the mighty ship Google is too big to sink? It hasn’t been a leader to date in AI and is reportedly playing catch up.

Tech Giants Want In on Electricity, Google Has a Foothold

by Leave a Comment

During the desperate days of the energy crisis in the 1970s, it looked as though the shortage was permanent and we would have to change the way we lived, worked and played to accommodate it.

In the end, it was technology that solved the crisis.

For fossil fuels, 3D seismic, horizontal drilling and fracking were used. For electricity, it was wind and solar, and better technology for making electricity with gas — a swing from burning it under boilers to burning it in aero-derivative turbines, essentially airplane engines on the ground.

A new energy shortage — this time confined to electricity — is in the making. There are a lot of people who think that, magically, the big tech companies, headed by Alphabet’s Google, will jump in and use their tech muscle to solve the crisis.

The fact is that the tech giants, including Google, Amazon, Microsoft, Apple and Meta, are highly interested in electricity because they depend on supplies for their voracious data centers. According to many experts, the electricity demand will increase exponentially as AI takes hold.

The tech giants are aware of this and have been busy as collaborators and innovators in the electric space. They want to ensure an adequate electricity supply and insist it is green and carbon-free.

Google has been a player in the energy field with its Nest Renew service. This year, it stepped up its participation by merging with OhmConnect to form Renew Home. It is what its CEO, Ben Brown, and others call a virtual power plant (VPP). These are favored by environmentalists and utilities.

A VPP collects or saves energy from the system without requiring additional generation. It can be hooking up solar panels and domestic batteries or plugging in and reversing the flow from an electric vehicle at night.

For Renew Home, the emphasis is definitely the home, Brown said in an interview.

For cash or other incentives (like rebates), participants cut their home consumption, managed by a smart meter so that air conditioning can be put up a few notches, washing machines are turned off, and an EV can be reversed to feed the grid.

Brown said that at present, Renew Home controls about 3 gigawatts of residential energy use —  a gigawatt is sometimes described as enough electricity to power San Francisco — and plans to expand that to 50 GW by 2030. All of it is already in the system and doesn’t require new lines, power plants or infrastructure.

“We are hooking up millions of customers,” he said, adding that Renew Home is cooperating with 100 utilities.

Fortunately, peak demand and the ability to save on home consumption coincide between 5 p.m. and 9 p.m.

There is no question that more electricity will be needed as the nation electrifies its transportation and its manufacturing — and especially as AI takes hold across the board.

Todd Snitchler, president of the Electric Power Supply Association, told the annual meeting of the U.S. Energy Association that a web search using ChatGPT uses nine times as much power as a routine Google search.

Google and the other four tech giants are in the electric supply space, but not in the way people expect. Renew Home is an example; although Google’s name isn’t directly connected, it is the driving force behind Renew Home.

Sidewalk Infrastructure Partners (SIP) has invested $100 million in Renew Home. Brown is a former Google executive, and Jonathan Winer is a co-CEO and cofounder of SIP.

As Jim Robb, the president of the North American Electric Reliability Corp., the congressionally mandated, not-for-profit supply watchdog, said recently on the TV show “White House Chronicle,” the expectation that Google will go out and build power plants is silly as they would face the same hurdles that electric utilities already face.

But Google is keenly interested in power supply, as are the other tech behemoths. The Economist reports they are talking to utilities and plant operators about partnering on new capacity.

Also, they are showing an interest in small modular reactors and are working with entrepreneurial power providers on building capacity, with the tech company taking the risk. Microsoft has signed a power-purchase agreement with Helion Energy, a fusion power developer.

Big tech is on the move in the electric space. It may even pull nuclear across the finish line.

Bill Gates and the Energy Research Dilemma

by 3 Comments

There is an idea that has been around for a long time, at least since the fall of 1973: All that stands between the United States and an abundant energy future is a lack of spending on research and development.

It is as though the Knights Templar could find the Holy Grail, if only the pope would commit just a few more resources to the hunt.

Tens of billions of dollars have been spent, many of them fruitlessly; and some advances have been made, not the least in the kind of drilling technology that enables us to drill miles below the sea floor in the Gulf of Mexico. (Oops!)

Much else has been researched and not come to market. Wind and solar have taken giant strides, but still require tax breaks and subsidies. Nuclear energy through nuclear fission has been researched, even as its deployment has slowed. Worldwide hundreds of billions of dollars have been spent on nuclear fusion with nothing to show for it. Other programs have gone by the board, from coal liquefaction to magneto hydrodynamics and ocean-thermal gradients.

The thing about energy research has been that there are many promising lines, but seldom a big success.

On Thursday, a new set of highly qualified persuaders came to Washington to exhort the government to increase energy research and development funding from $5 billion to $16 billion a year, and to set up new organizations to channel and manage basic research on energy.

Some of the nation’s industrial savants, including Bill Gates late of Microsoft, Jeff Immelt of General Electric and Ursula Burns of Xerox, appeared at a press conference here as members of the American Energy Innovation Council. The chairman of the group, Chad Holliday of Bank of America, told the press: “Up until now energy investments have gotten short shrift.”

That is debatable. The problem with energy research has not been that it has been shortchanged, but that it has often been directed at the wrong thing; it has often been diluted or spread out for political purposes. Farmers want ethanol research, coal states want carbon management, and the populous Eastern states want carbon-free energy — so long as it is not nuclear.

The group of industry captains is not looking at the political, social and economic divides that have negated so many past endeavors. Just when the nuclear industry was ready to enter its long-expected renaissance in the 1990s, it was broadsided by new gas turbines. If the carbon in coal can be safely sequestered, does that solve the environmental problems of ripping it out of the ground?

R&D always produces something of interest and often of value, but not always what it was directed toward. At the press conference, Xerox’s Burns said that innovation needed to be managed, and that the CEOs of the group knew that from experience.

Actually, the experience of Xerox itself may belie that. The original copying machine technology nearly perished for want of sponsorship and was finally saved by not-for-profit Battelle Laboratories. Later, when many of the innovations that made the rise of Microsoft, Apple and Cisco possible were developed at Xerox’s California computer laboratories, the company did not know what to do with them. But Bill Gates did. These two should talk.

The great Bell Labs produced optic fiber and the transistor, but did nothing with them. Management is a lovely business when it controls but in so doing, it stifles.

If you want innovation, first get rid of the managers; second, get on bended knee before the bankers.

A new attitude toward energy is needed, but first it is a good idea to know where we want to go.

With the catastrophe in the Gulf, our energy future is again in flux. The trusted has become dangerous, and the dangerous may again be trusted. –For the Hearst-New York Times Syndicate