Prometheus Project - Developer Mode
The Prometheus Project - Switching to developer mode requires administrator authentication. “Of course.” Yoo Jae-won entered the password only he knew into the message box that appeared on his smartphone.
Switch complete. Initiating unlimited advanced computation.
There was no tedious back-and-forth about conditions or re-explaining the core of the question. Developer-mode Gold already had vast quantities of data and countless variables pre-configured. What’s more, it had been deliberately evolved with the expectation that it would exceed routine support duties, which meant it could now make reasonably sophisticated value judgments on its own.
Ding!
The message arrived instantly from Yeong-sik, who oversaw ID Group’s cloud center. While Gilbert had designed the software, Yeong-sik was responsible for the entire hardware backbone. The moment he saw a specific process demanding a massive spike in computational resources, he had fired off an ID Talk to Yoo Jae-won.
“I asked it a tough question in developer mode.”
“Sorry, I should have warned you. Something just popped into my head on the spur of the moment.”
One of machine learning’s fatal flaws was the infinite loop. If circular logic kicked in or a routine resembling a logic bomb appeared, the consequences could be disastrous. With an astronomical number of operations running in parallel, any delay in noticing the problem meant nothing but wasted computing power.
“Give it exclusive access to ten exaflops for just one day.”
ID Group’s cloud infrastructure had so much headroom that peeling off ten exaflops for a single process barely made a dent.
“The answer will probably be obvious anyway.”
Everyday assistance was already well within the current Gold’s capabilities. It was actually superior to humans at repetitive tasks. Organizing photo libraries was a perfect example. If someone had dumped thousands or tens of thousands of unorganized pictures into a folder, Gold’s help was ideal. It could recognize image content, sort photos into folders, and apply tags with remarkable speed and accuracy. When it encountered ambiguous images it would ask the user for clarification, and every honest answer improved its learning model, raising future classification efficiency.
Complex strategic questions, however—such as where Lightning Bolt should locate its mass-production plant—lay beyond its current reach. As Gold’s creator, Yoo Jae-won already had a rough idea of the quality he could expect. He had even built a separate intelligence-testing suite usable for both strong and weak AI. The publicly released Gold scored 86 on the test; the developer-mode version he was using now scored 92. Think of it like human IQ. The numbers alone showed there was still a long road ahead.
Still, President Bolt had promised to select the optimal site and submit a report soon. Comparing that document with Gold’s output would let Yoo Jae-won gauge the hidden growth of the developer-mode version. That was why he had issued the command.
The next day.
After completing his usual crisp morning routine, Yoo Jae-won entered his study, sat at his desk, and logged into both his smartphone and computer. Staying logged in was convenient, but terrible for security; therefore, entering his password every single time had become habit.
Chairman Yoo Jae-won, welcome.
The new addition was Gold’s personalized greeting.
“Good morning. What’s the news lineup today?”
There are thirty-seven news items you should review, Chairman.
The conversation between Yoo Jae-won and Gold flowed so naturally that a casual listener might have mistaken it for a talk with Chief of Staff Kim Dae-seok. The results, however, were still questionable. Compared to the tightly curated morning brief Kim Dae-seok delivered every day, Gold’s selection was a clear defeat. Who on earth needed to read thirty-seven separate news items before breakfast?
The screen Gold presented was a chaotic mix—everything from rising Korean housing prices to France’s immigration problems. Kim Dae-seok’s scrapbook, by contrast, was always concise and on-target.
There were logical reasons behind Gold’s choices. The Korean housing article was a follow-up to one Yoo Jae-won had clicked previously, and the French immigration piece was currently the top-clicked story on Nextcom France. Yet none of them were truly vital to the man who would actually read them.
“Starting tomorrow, keep the news list under ten items, please.”
Understood. I will adjust the news scrap limit to ten items.
If the cap had existed from the start, today’s overload would never have happened. Still, imposing an arbitrary limit at the beginning would have constrained Gold’s logical growth, so he had left it unrestricted. Now that the experiment had produced near-useless results, trimming the number was the correct—if regrettable—move.
That didn’t mean every article Gold chose was worthless. The surge in Korean housing prices, in particular, was an issue Yoo Jae-won had been watching closely.
The article made a big fuss about prices having risen forty percent since the early days of the Roh Moo-hyun administration, but by Yoo Jae-won’s standards the situation was not yet critical. An apartment in Gangnam could still be bought comfortably for the high 300-million-won range.
Besides, the Roh Moo-hyun administration was tackling real estate far more proactively than in his previous timeline. Back then they had tried to handle everything at once—historical liquidation, private-school reform, and more—only to trigger fierce backlash and fail across the board. This time, having successfully completed private-school reform by embedding public-interest directors on foundation boards, the administration was concentrating its firepower on real estate.
Exactly how they would address it, however, had not yet been decided. The chance of a comprehensive real-estate tax appearing was high, but it was still too early to be certain. The conditions were markedly different now. Most importantly, the volume of liquid capital circulating in Korea had grown enormously compared to the past. The swift recovery from the IMF crisis was the primary reason. During the IMF period money had dried up completely; now there was so much liquidity that inflation had become a genuine concern.
Fortunately, the stock market and financial products such as funds had matured far beyond their previous levels, absorbing much of the excess capital that might otherwise have flooded real estate. Still, there were limits. Korea’s financial markets had not yet grown large enough to soak up all the wandering liquidity.
“It’s not as if there are no solutions.”
The answer was to list several ID Group subsidiaries. Creating legitimate outlets for massive capital inflows would suppress the real-estate bubble as much as possible.
The companies he had in mind for domestic listing were ID Electronics and ID Display. Both were super-blue-chip giants already dominating their IT sectors. Electronics held more than half the global market in home appliances, memory semiconductors, and non-memory semiconductors alike. A listing would draw not only Korean liquidity but investors from around the world eager to buy shares.
ID Display was no different. Bordeaux TV had become the global standard for flat-screen televisions, and ID Display’s monitor panels for computers and smartphones were universally recognized as top-tier. Chinese firms were chasing aggressively, but when facility investments were measured in trillions of won, catching up was still a distant dream.
“There’s also Gigastudy.”
The undisputed leader in online education boasted equally impressive performance. It had begun with modest 100-day CSAT prep courses but now offered a vast curriculum covering middle-school classes, civil-service exams, national certifications, and more. Foreign-language instruction could not be overlooked either. Having secured millions of paying users inside Korea alone, Gigastudy was expanding vigorously overseas—sometimes by launching direct services, sometimes by exporting the platform itself.
Having embraced interactive internet lectures since the 1990s, Gigastudy had kept pace with the smartphone era by developing a full mobile platform. More than that, it had continuously refined the service through feedback from instructors and students alike, adding new features until it became the world’s premier online education system. Korean analysts were already predicting that a listing would easily push its market capitalization past one trillion won.
If these ID Group subsidiaries were brought to the Korean stock market, they would instantly absorb the excess liquidity drifting through the economy. The upward momentum in real-estate prices would vanish overnight.
Yoo Jae-won disliked IPOs, yet he was willing to endure the inconvenience because he feared a property bubble. The reason was simple: when enormous capital became trapped in real estate—which produced no added value—everyone except a tiny handful of “building owners” suffered. It was fundamentally irrational for more than half of a month’s hard-earned salary to disappear into loan repayments or rent. Likewise, a society in which decades of work still could not secure a home for most citizens was unhealthy for everyone.
Conversely, if that same liquidity were channeled into technological progress, there could be no better outcome.
Of course, investment in technology was not automatically virtuous. The fallout from bursting bubbles like the IT bubble was devastating, and swindlers were legion. In the late nineties, companies with “techni-network” in their names had attracted blind investment; now huge sums were pouring into bio-ventures on the back of Hwang Woo-suk’s stem-cell claims. Even his great-uncle and other relatives had shown him lists of such firms and asked whether investing would be safe. Since every name on those lists was completely unknown to him, Yoo Jae-won had answered with flat rejection. Even so, such investments were still far better than burying money in land.
“Tsk. Then why are they only watching the thorium reactor?”
While scanning Korean news, Yoo Jae-won’s thoughts abruptly shifted to his own business. He pulled up a document from the company intranet titled Thorium Reactor Purchase Negotiation Status.
United States, United Kingdom, France, Germany, Japan, Italy, Canada, Russia. The G8 advanced economies are all engaged in negotiations without exception. Additional parties include energy-hungry nations such as China, Korea, Brazil, and India. North Korea is also participating.
These are formidable countries; contracts could be signed at any moment. Nevertheless, not a single formal agreement has been reached.
“Hey, Gold. Why aren’t the thorium reactors selling? Or am I simply being impatient?”
That is a difficult question. Would you like me to calculate it using advanced computation?
“No. One query is already running—let’s not add another. I’ll pass.”
Yoo Jae-won let out a small sigh. The question he had posed yesterday—where Lightning Bolt should build its mass-production vehicle plant for maximum benefit to both himself and ID Group—was still being processed. Even with supercomputer-level resources it would take several more hours to produce an answer.
At moments like this he inevitably remembered the AIs from his previous life. Even the publicly released AI secretaries of that era had been incomparably more advanced than the current Gold. The undisputed champion had been Google’s Golden Silver, a monstrous system that combined silicon-based cloud architecture with quantum CPUs. Its power was so overwhelming that ordinary users could never access its full capability. Yoo Jae-won, however, had proven his worth and earned direct-access rights granted to only a handful of people. Using that access he had completed his machine-psychology module. The performance he had witnessed then remained vividly etched in his memory.
“At least the most enthusiastic buyer so far is North Korea?”
A report from ID Westinghouse’s Thorium Reactor Business Unit appeared on his monitor. The old reactor division of the once-dying Westinghouse had been completely dismantled and replaced by the new thorium unit. The bloated bureaucracy was being restructured into the lean, team-based organization ID Group favored. Resistance had been fierce—phrases like “occupation force” were tossed around freely—but there was no way to stop the change. In any case, the new organization was up and running, steadily producing reports.
As expected, many countries had been tempted by the 1+1 promotional event, and North Korea was the most aggressive of all. The report noted that the price—half that of conventional reactors—and the use of stable thorium were major advantages. Consequently, North Korean observers had been diligent visitors to the 2-gigawatt plant under construction in the Mojave Desert and made frequent trips to the experimental reactor in Waterville, meticulously checking data.
Yet not a single formal contract request had arrived. Even with such an unprecedented promotion, the situation remained unchanged. It seemed orders would only begin after the Mojave thorium reactor proved it could operate reliably at full scale.
Ding!
While Yoo Jae-won was still contemplating the thorium reactor, an alarm sounded.
“Ah, it’s finally here.”
The sender was President Bolt of Lightning Bolt. The document title was Mass Production Master Plan—exactly the report Yoo Jae-won had been waiting for these past several days.