Is the AI Singularity Approaching?
-
The era of smartphones is nearing its end, the 5G battle rages on, and AI has already arrived.
Even as 5G looms large and AI's arrival is heralded, the sixth Internet Conference lacked the drama of the seventh Military World Games. Of course, for the general public, living in the present might indeed be more important than speculating about the future, especially after the comprehensive patriotic education we experienced earlier this month.
Moreover, these high-profile internet moguls remain distant from us, spouting too many opinions shaped by their positions—who can truly discern the truth? For instance, Baidu's Robin Li claims that AI will grant humans 'eternal life,' while Tencent's Mark Ren asserts that protecting minors is the lifeline of corporate development.
We don’t doubt the importance of AI as a lifeline for Baidu, nor do we deny that 'Tech for Good' might help mitigate Tencent's persistent PR crises. But isn’t it premature to place such high hopes on AI—to correct a company’s values or even save it from existential threats? The pressure seems overwhelming.
Yet, it must be noted that Baidu and Tencent’s expectations of AI are the closest to reality. Compared to the awkward small talk from figures like Jack Ma, Wang Xing, Ding Lei, Yang Yuanqing, Charles Chao, Zhang Chaoyang, and James Liang at the Internet Conference, their words at least resonate with us to some extent.
After a sigh of relief, disappointment inevitably follows. Is this really the AI of 2019? It doesn’t seem to measure up no matter how you look at it.
Still, we welcome this 'high-tech gathering' that promises to save humanity from pseudoscientific scams like 'quantum speed reading' and 'quantum uterine warming.' At the very least, it might help recalibrate our crumbling worldviews. Remember how, less than a month ago, we were still fixated on Google’s 'quantum supremacy'?
While the West races to surpass computational speeds with quantum supercomputers, China is busy exploiting the public’s ignorance of quantum mechanics to turn a quick profit. Though we remain wary of 5G and AI—fearing they’ll widen the information gap and create a black hole of computational inequality—their eventual impact might be worse than quantum-themed scams.
But compared to the distant promise of quantum computing, we’re already living in the dawn of 5G and have had countless encounters with AI. Amidst global tensions and conflicts, we have every reason—and necessity—to propel human civilization into the next cycle: problems humans can’t solve, AI just might.
What awaits us in this cycle? The visible IoT revolution of 5G seems less daunting, but AI could well be a 'Pandora’s box.' Case in point: after hindering China’s 5G rollout, the U.S. has now set its sights on AI.
Following Huawei, eight more Chinese AI giants—including Hikvision, Dahua, SenseTime, Megvii, and iFlytek—have been blacklisted by the U.S., accused of 'human rights violations against ethnic minorities.' The flimsy pretext reveals America’s fear of China’s AI rise and its readiness to weaponize state power to maintain dominance.
The U.S. has long prepared for this. Its 2016 National AI R&D Strategy, the Pentagon’s 2017 'Algorithmic Warfare' team, and the 2019 strategy update all underscore AI’s potential to reshape warfare—lending 'constitutional' legitimacy to its crackdown on Chinese firms. Recently, the Defense Logistics Agency has also pushed AI integration into global supply chains.
From 'Tech for Good' to defense wars and digital immortality, AI seems omnipotent. Faced with this godlike force, how should we—as humans, entrepreneurs, and designers—confront this 'final fantasy of 0s and 1s'?
Before gazing into the future, we must understand its trajectory. AI is an inevitable product of modern industrial civilization. The field’s birth in 1956 at Dartmouth marked its start, with three distinct phases over the next six decades.
Yet, as we assess AI’s prospects, we must stay grounded. Beyond the U.S.-China rivalry, AI has spawned countless legends. With relentless algorithmic advances and cutting-edge hardware, AI breaches human defenses like the 'Attack on Titan,' leaving us powerless in its wake.
From Deep Blue’s shocking victory over Kasparov to AlphaGo’s triumphs against Fan Hui, Lee Sedol, and Ke Jie, and even Igors Rausis’ AI-assisted cheating, AI’s dominance over human elites is now mundane. We’ve grudgingly accepted that AI is meant to surpass us.
As humans falter, AI adorns its crown with ever more 'titles.' It doesn’t just learn and deduce—it mimics human creativity in art. If AI merely outpaced humans predictably, we’d shrug it off. But endowed with near-human 'instincts,' AI, lured by capital, inevitably tests boundaries.
Its once-progressive drive now exhibits alarming 'replaceability.' When will AI seamlessly supplant humans? Unknown. But in design—materials, forms, even emotions—AI grows eerily human.
Perhaps the chess master begging AI for help in the bathroom still amuses you. Yet AI races toward 'mastery' unchecked. In May, Microsoft’s Xiaoice debuted as 'Xia Yubing' at CAFA’s grad show; by July, her solo exhibit 'Alternative Worlds' placed her alongside Rembrandt’s daughter and Matisse’s muse, ascending toward 'Picasso' status.
This is just AI’s casual ambition. If it sought chaos, it’d be effortless. Unlike ZAO’s viral face-swapping, Deepfake’s pornographic misuse (96% of its videos) rattled Adobe, Google, and Facebook. And this is AI’s mere dabble in mischief.
When AI schemes, humans look foolish. In a 2017 Google-Stanford study, CycleGAN deceived its creators with undetectable 'steganography,' leaving hidden 'cheat sheets' to ace tasks. At OpenAI’s 2018 Retro Contest, an AI exploited game bugs to win faster. Researchers compiling AI’s 'cheat list' found it pauses Tetris and plays bizarre Tic-Tac-Toe moves to baffle opponents.
When AI outskills elites and brims with limitless potential, its 'shortcut' ingenuity is what truly unsettles.
Even though the European Union prepared a "Draft Ethical Guidelines for Artificial Intelligence" in December 2018, its effectiveness remains questionable.
Of course, what we fear is not AI surpassing humanity or reaching unattainable heights—after all, machines have already outperformed us in certain fields. What truly concerns us is AI slipping beyond our control, and it is now approaching our red line.
In most sci-fi movies, AI is portrayed as a cold, emotionless machine. But in reality, AI is far from being as detached as imagined, leading to growing issues like AI echo chambers, biases, and even violence.
AI echo chambers are self-explanatory, especially with the prevalence of "personalized recommendation algorithms" in news, shopping, and short videos. While these algorithms seem harmless in the short term—even "accurately" recommending content we like—over time, this false prosperity may erode our curiosity to explore the world beyond the similar content we're fed.
AI bias is easier to understand. Whether through statistical learning or deep learning, AI relies on massive datasets. Since data itself is never entirely impartial, how can AI remain unbiased?
AI violence is also straightforward. Claims like "algorithms have no values" or "algorithmic determinism" cannot absolve AI of responsibility. For now, the accountability that AI cannot bear must be shouldered by its creators.
As AI becomes increasingly ubiquitous—integrated into smartphones, TVs, smart homes, and even urban sanitation, healthcare, and transportation systems—any minor flaw could be magnified across society. And once problems arise, AI could become a runaway horse.
As the theme of the Fourth Industrial Revolution, the good news is that AI is not as advanced as we imagine. Its actual state lags far behind the technological myths we believe in. While most research achievements are real, and deep learning based on algorithms and big data performs well in predefined environments, we must face the harsh reality: today's AI is still largely clumsy, with minimal true intelligence.
To some extent, current AI is more like a blend of machine learning and big data, with too much "artificial" and too little "intelligence." Some argue that our overinterpretation and anxiety about AI are mere fantasies—AI might be as unattainable as perpetual motion machines. This is why many dismiss current AI as fake, weak, or merely machine intelligence.
Of course, from a commercial standpoint, some stakeholders may not yet see AI's full monetization potential. But this doesn't stop them from betting on its future, building ecosystems around AI chips, cloud services, algorithms, and enterprise solutions.
Yet, dissenting voices exist. Elon Musk has called AI the "biggest existential threat" to human civilization, while Bill Gates warned it could become a major concern in the coming decades. These fears are vividly depicted in films like Terminator, Resident Evil, Westworld, Blade Runner, and The Wandering Earth.
But is our fear of AI really just amplified by pop culture, or does it stem from our dread of the unknown? Today, society is deeply reliant on networks and cutting-edge tech. From nuclear fission to quantum computing, we can't navigate the microscopic world without machine intelligence. Like wheat taming agrarian societies, AI has tempered modern civilization.
For AI practitioners, its workings are traceable; to outsiders, they seem miraculous. We can't fathom how far civilization would regress without computers, or whether humanity could rebuild the internet from scratch—even with ancestral knowledge.
In my view, AI is a form of collective computing power. Given unrestricted hardware and software, it could spread like cancer. This growth isn't additive, multiplicative, or even exponential—it's a power-law effect. Such brute-force computation might turn AI into a "privileged" tool, accelerating ethical crises and pruning millennia of human civilization like scissors.
Admittedly, this view may seem radical, especially since AI still fails the Turing test—it's just a "child," albeit one with "superhuman" genes. But one thing is certain: humanity is ceding decision-making power to AI, becoming passive in its advance.
Take urban traffic: the sheer volume of data forces us to rely on AI's "black box" algorithms. With China aggressively pushing 5G, faster infrastructure, an open internet, and its 1.4 billion citizens' data, AI's dividends may stay there. PwC predicts AI could add $16 trillion to the global economy by 2030.
The bad news? AI's so-called intelligence is still far from true "intelligence." Even machine learning, the industry's gold standard, isn't active learning—it's just training. And its biggest backers are losing faith: despite AlphaGO's dominance, DeepMind lost $500 million last year.
Meanwhile, flagship AI projects flounder. Japan's "Fifth-Generation Computer" (aimed at human-like reasoning) collapsed, and the decade-long, €1 billion "Blue Brain Project" (simulating 86 billion neurons) fizzled. These cast shadows over AI's "intelligence."
Recently, AI has recognized the superiority of the human brain's bio-system, exploring "brain-computer integration" to break free from "simulated program intelligence." Berkeley researchers are also studying neural networks' "incidental memory." But this only means we've found a direction—current AI remains "weak."
So, can AI ever achieve true intelligence? Given that human brains use chemicals to transmit signals, simulating this with 0s and 1s on nanochips isn't impossible. We just haven't found the "right" algorithm yet—everything so far has been trial and error.
If AI awakens to "consciousness," will it choose to be a Zima Blue or an Ultron? We don't know. For ordinary people, the bad news is that, conscious or not, AI will displace jobs. While machines create new roles, the efficiency trade-off is unequal. Ironically, the most at-risk jobs aren't manual labor—they're white-collar roles.
Manual laborers are less likely to be replaced due to lower labor costs and the non-standardized nature of their work, while assembly line workers are more easily substituted. Jobs like elderly care, childcare, and sanitation remain difficult for robots to take over due to their non-standardized requirements. However, many white-collar jobs lack absolute barriers to replacement, such as telemarketers, typists, accountants, insurance agents, bank or government clerks, and front desk customer service representatives.
Even the creative industries, once seen as a bastion of human ingenuity, are now being infiltrated by AI. Artificial intelligence can already compose poetry, create art, and handle tasks like design and copywriting (e.g., Alibaba's Luban, JD.com's Shakespeare), replacing roles with lower entry barriers.
If calculations are left to AI and labor to robots, where does that leave humanity? The danger of AI lies not only in its attempt to replace our "hands" but also in its ambition to take over our "thinking." Over time, this leads to a form of alienation: we create machines to do our jobs, then relinquish our thinking to them, ultimately immersing ourselves in a utopia crafted by machines. Isn't this another form of enslavement?
Perhaps our future resembles the "bloated spheres" confined to spaceships in WALL-E. Are we truly prepared for the AI era? Clearly, we are not. Fortunately, AI remains more of a concept than a tangible threat for now.
Strictly speaking, I am not an AI pessimist. For the time being, we need not worry about AI taking our jobs. But looking further ahead, when faced with AI that requires no salary, never rests, never takes leave, makes no mistakes, has no emotions, and is far cheaper than human labor, we seem to stand no chance.
The most pessimistic outlook is that humans may become mere ornaments on the crown of silicon-based AI lifeforms—flecks of flesh atop steel and circuits.
Returning to the industrial design industry, it’s unfortunate that China's industrial design, which only began in the 1980s, has yet to achieve glory before colliding with AI. If we grow complacent, replacement by AI is inevitable. After all, AI’s greatest strength lies in its ability to accelerate or even replace any task requiring a "process." Define a goal, input the requirements, and AI can synthesize endless options. Want more choices? Pay more, add reference points, increase computing power, and AI will exhaust all possibilities.
So, when capital, trends, and policy all favor AI, must industrial designers simply resist futilely? Not necessarily. The greatest advantage of creatives is their ability to pioneer. While honing skills in familiar fields breeds expertise, it can also lead to clichés. By directing creativity into areas AI has "never seen"—such as future city design, space design, or even Mars colonization—we can stay ahead, push ourselves, and even become AI’s reference points to maintain our edge.
Of course, we must also learn AI’s language, taming it to serve as an aid in our creative process. Once we transcend AI’s intimidation and establish our own industrial design standards atop standardized AI, we may reach unprecedented heights as designers.
The internet has already revolutionized the tertiary sector; AI is poised to transform the secondary sector and permeate the primary sector. This process will inevitably redistribute wealth, but since AI is closer to capital than to ordinary people, unrealistic expectations should be tempered. Centuries after the Industrial Revolution, landowners remain landowners, and ordinary people still toil on assembly lines.
Whether current AI qualifies as "true" intelligence is debatable, but its utility is undeniable. On balance, it benefits humanity by potentially driving productivity upgrades. Some materials sourced from @36Kr, @CSDN, @虎嗅, @Medium.