2026年7月31日 星期五

中國經濟成長創多年來新低(1/2)

Recently The New York Times reported the following:

China’s Economy Grows at Slowest Pace in Years (1/2)

Economic growth of 4.3 percent in the second quarter, versus the same period last year, reflected a broad slump outside of the country’s export-oriented manufacturing might.

The NYT -By Alexandra Stevenson and Murphy Zhao - Reporting from Hong Kong

July 15, 2026

Updated 2:38 a.m. ET

China’s economy last quarter grew at the slowest rate in three years, reflecting a broader slump that the country’s leaders signaled earlier this year when they set the lowest growth target in more than three decades.

On Wednesday, that National Bureau of Statistics said that the economy expanded by 4.3 percent in the second quarter, compared to a year ago, down from a 5 percent pace in the first quarter and short of economists’ expectations.

Although China’s factories are churning out chips and electric cars to supply a global boom in artificial intelligence and energy-saving products, many Chinese people are feeling squeezed at home.

A long-running property crisis has no end in sight, with steep declines in construction dragging down economic growth. Jobs outside of factories are hard to come by and paychecks are not growing. Retail sales of consumer goods have been choppy. They fell in May, for the first time since the end of Covid-19 lockdowns in late 2022, before recovering somewhat in June.

That is in stark contrast to China’s relentless strength in manufacturing and trade, with a government report released on Tuesday showing China’s exports surging by 27 percent in June compared with a year earlier, driven by shipments of chips, batteries and cars. China’s trade surplus in June, at more than $125 billion, was second largest on record.

In other words, China’s mighty export machine is masking weaknesses elsewhere.

“You get this A.I. boom, which is a global thing, and China is part of the leading nations on the frontier,” said Yu Song, the chief China economist at UBS Securities. “Without this, China’s economy would be in a much worse state.”

When measured on a quarter-to-quarter basis, China’s economy expanded by only 0.9 percent in the second quarter. When projected out for a year, the second-quarter data implies that the economy was growing at an annual rate of 3.6 percent, sharply down from a pace of more than 6 percent in the first quarter.

The second-quarter annualized rate also missed official targets. Shortcomings in China’s economic growth drivers prompted the ruling Communist Party earlier this year to set the lowest annual growth target in decades, with a goal of between 4.5 percent and 5 percent this year.

Stepping back to consider the trends in the first half of the year, economists say that China’s manufacturing and export prowess, however robust, cannot carry growth on its own.

Industrial production rose by 5.4 percent in the first six months of the year, versus the same period last year. High-tech manufacturing rose by more than 13 percent over that period. But fixed asset investment — which includes infrastructure, property construction and manufacturing — fell by 5.7 percent. Real estate development dropped 18 percent.

The value of China’s exports surged by more than 20 percent in the first half. But consumer spending, which a Moody’s Analytics report said “remains the economy’s weakest link,” faltered. Retail sales of consumer goods increased by 1.3 percent over the first half of the year.

The effects of the war in Iran have pinched Chinese households, with rising fuel prices prompting them to drive and fly less, at a time when many were already worried about the economy and choosing to save more.

China has softened the blow of rising fuel costs by controlling the price at the pump, but the cost of filling up for drivers is still double-digit percentages higher than a year ago.

One silver lining, economists said, was that rising fuel prices started to feed through to broader inflation in the quarter, reversing a problem that China has struggled to shake: more than three years of a broad-based decline in prices. Such deflation tends to chill spending, with consumers putting off purchases in expectation that prices will be lower in the future.

China’s gross domestic product deflator, a broad measure of prices across the economy, was negative in 13 of the past 14 quarters — the most protracted slump on record. But it turned positive in the second quarter.

(to be continued)

Translation

中國經濟成長創多年來新低(1/2

 第二季經濟成長4.3%,與去年同期相比,反映出除出口導向製造業以外的整個經濟領域普遍下滑

中國經濟上季成長創三年新低,反映出更廣泛的經濟下滑趨勢。今年早些時候,中國領導人設定了三十多年來的最低成長目標,也預示了這一趨勢。

週三,國家統計局公佈的數據顯示,第二季中國經濟年增4.3%,低於第一季的5%,也低於經濟學家的預期。

儘管中國工廠正源源不絕地生產晶片和電動車,以滿足全球人工智能和節能產品的需求,但許多中國人卻在國內感到生活拮据。

長期存在的房地產危機仍看不到盡頭,建築業的急劇下滑拖累了經濟成長。工廠以外的工作機會難得,薪資成長也停滯不前。消費品零售額波動不定。 5月份,自2022年底新冠疫情封鎖措施解除以來,零售額首次出現下滑,6月略有回升。

這與中國製造業和貿易的強勁勢頭形成鮮明對比。週二發佈的一份政府報告顯示,受晶片、電池和汽車出貨量的推動,中國6月出口額年增27% 6月中國貿易順差超過1,250億美元,創史上第二高。

換句話說,中國強大的出口機器遮掩了其他方面的疲軟。

瑞銀證券首席中國經濟學家Yu Song表示:“人工智能的蓬勃發展是全球性的,而中國是走在前端的領先國家之一。如果沒有人工智能,中國經濟的狀況會糟糕得多。”

按季度環比計算,中國經濟第二季僅成長0.9%。若將第二季的數據推算至一年,則經濟年化成長率為3.6%,較第一季超過6%的成長速度大幅下降。

第二季的年化成長率也低於官方目標。由於中國經濟成長動力不足,執政的中國共產黨在今年稍早設定了數十年來最低的年度成長目標,將今年的成長率目標設定在4.5%5%之間。

回顧上半年的發展趨勢,經濟學家指出,儘管中國的製造業和出口實力強勁,但僅靠這些優勢無法支撐經濟成長。

相比去年同期,今年上半年,工業生產年增5.4%。高科技製造業年增超過13%。但固定資產投資(包括基礎建設、房地產建設和製造業)下降5.7%。房地產開發下降18%

中國上半年出口額成長超過20%。但消費支出卻步履維艱 - 穆迪分析公司一份報告稱,消費支出「仍然是經濟最薄弱的一環」。上半年,消費品零售額成長1.3%

伊朗戰爭的影響給中國家庭帶來了壓力,不斷上漲的燃料價格促使他們減少駕車和乘坐飛機。當前許多人已經對經濟狀況感到擔憂,並選擇增加儲蓄。

中國透過控制油價緩解了燃油成本上漲的衝擊,但目前駕駛者的入油成本仍比一年前高出兩位數百份比。

經濟學家指出,有一線希望在於燃油價格上漲在本季開始傳導至更廣泛的通膨,扭轉了中國長期以來難以擺脫的通縮局面:三年多來物價普遍下跌。這種通貨緊縮往往會抑制消費,消費者會延後購買,預期未來價格會更低。

中國的國內生產總值平減指數(衡量經濟整體物價水準的指標)在過去14季中有13季為負值,創下歷史最長的低迷期。但該指數在第二季轉正。

(待續)

2026年7月30日 星期四

SoftBank Group CEO Masayoshi Son: "The Landscape Will Change with the Spread of AI Agents"

Recently NHK News on-line reported the following:

ソフトバンクG 孫社長 “AIエージェント普及で景色変わる”

202671415:34

(202671420:59更新)

生成AI・人工知能

ソフトバンクグループの孫正義社長は14日の講演で、2040年には自律的に仕事を行う「AIエージェント」の普及が見込まれるとして、稼働に必要な電力やデータセンターといったインフラ投資を強化していく考えを示しました。

ソフトバンクグループの孫正義社長は、14日、取引先などを集めたイベントで講演しました。

この中で孫社長は、AIの未来像について、2040年には人に代わって自律的に仕事を行う「AIエージェント」が広く普及していることが見込まれるとした上で、「AIエージェントが世界に100兆個誕生する。AIエージェントを宿したロボットが肉体労働の中心になるだろう。世の中の景色は完全に変わる」と述べました。

その上で、膨大なAIを動かすためにはデータセンターや電力、半導体などのAIインフラへの投資が世界で年間5兆ドル、日本円で800兆円規模で必要になるという試算を示し、「非常に大きな金額だが、2040年には世界のGDP20%をAIが占めると見ている。日本にとどまるだけでなくアメリカを中心に攻めていかないといけない」と述べ、会社としてAIインフラ投資を強化していく考えを示しました。

AI組み込んだロボットアーム披露 フィジカルAIの開発で

ソフトバンクグループでは、AIが自律的に判断してロボットなどを動かす「フィジカルAI」の開発にも力を入れていて、会場では、AIと組み合わせた最新のロボットアームが実演されました。

このロボットアームは、子会社のソフトバンクと産業用ロボットの大手メーカー「安川電機」が共同で開発しました。

従来のロボットアームでは、前もって配置が決まっている上、形も安定したモノしか持ち上げることができませんでしたが、AIがカメラで形を認識することで布やひもなどもつかめるようになったということです。

会場では実際にロボットアームが、電気ケーブルをつかんで箱の中に収納する様子が披露されました。

会社では、これらのロボットアームを活用できれば、オフィスでの在庫の補充や掃除のほか、家庭で洗濯物をたたむなど工場以外での活用も期待できるとしています。

ソフトバンクの先端AIプラットフォーム統括部の山科瞬統括部長は「人材不足が課題になる中で、これまで人しかできなかった作業をフィジカルAIが実現できるようになれば、労働力を補うことができる。今後フィジカルAIができることを増やしていく必要がある」と話しています。

Translation

SoftBank Group CEO Masayoshi Son: "The Landscape Will Change with the Spread of AI Agents"

July 14, 2026, 3:34 PM

(Updated July 14, 2026, 8:59 PM)

Generative AI/Artificial Intelligence

SoftBank Group CEO Masayoshi Son (孫正義) stated in a speech on the 14th that he expected "AI agents" that could perform work autonomously would be widespread by 2040, and indicated his intention to strengthen investment in infrastructure such as power and data centers necessary for their operation.

SoftBank Group CEO Masayoshi Son gave a speech at an event attended by business partners and others on the 14th.

In his speech, CEO Son spoke about the future of AI, stating that "AI agents" that could perform work autonomously in place of humans were expected to be widely used by 2040. He added that "100 trillion AI agents will be created worldwide. Robots equipped with AI agents will become central to manual labor. The landscape of the world will completely change."

Furthermore, he presented an estimate that running a massive AI system would require an annual investment of $5 trillion globally, or approximately 800 trillion yen, in AI infrastructure such as data centers, power, and semiconductors. He stated, "This is a very large sum, but we predict that AI will account for 20% of global GDP by 2040. We must not only stay in Japan, we need to aggressively target markets, primarily focusing on the United States.," indicating the company's intention to strengthen its investment in AI infrastructure.

AI-Integrated Robot Arm Unveiled: Development of Physical AI

At SoftBank Group, they're also putting a lot of effort into developing 'Physical AI,' where AI autonomously made decisions to operate robots etc. At the venue, the latest robot arm combined with AI was demonstrated.

This robot arm was jointly developed by SoftBank, a subsidiary, together with Yaskawa Electric, a leading manufacturer of industrial robots.

Conventional robot arms could only pick up objects with predetermined positions and stable shapes, but by using AI to recognize shapes with a camera, it could now grasp cloth, strings, and other materials.

At the venue, a demonstration was given showing a robotic arm grasping electrical cables and storing them in a box.

The company expected that these robotic arms could be used not only for replenishing inventory and cleaning in offices, but also for tasks outside of factories, such as folding laundry at home.

Shun Yamashina (山科瞬), General Manager of SoftBank's Advanced AI Platform Coordination Department, said, "With labor shortages becoming a challenge, if physical AI can perform tasks that only humans could do before, it can supplement the workforce. We need to increase the capabilities of physical AI from now on."

So, conventional robot arms after equipped with AI to recognize shapes through a camera could now grasp cloth, strings, and other materials. In future it could also be designed for tasks outside of factories. SoftBank predicts that massive AI system will require an annual investment of $5 trillion globally, and will account for 20% of global GDP by 2040. Apparently, physical AI will be more and more popular soon.

2026年7月29日 星期三

Ukraine's Civilian Casualties Exceed 2,200 in June: Highest in Four Years - UN

 Recently NHK News on-line reported the following:

6月のウクライナ民間人死傷者2200人超 過去4年で最多 国連

20267158:33

ウクライナ情勢

国連の人権監視団は14日、ロシアによる軍事侵攻が続くウクライナについて、先月の民間人の死傷者は2200人を超え、侵攻直後の20224月以降、月別で最も多くなったと明らかにしました。

国連ウクライナ人権監視団は14日、ロシア軍の攻撃によるウクライナの民間人の死傷者は先月、少なくとも2283人だったと明らかにしました。

内訳は死者が293人、けが人が1990人で、侵攻直後の20224月以降、月別で最も多くなったとしています。

死傷者が増加した背景について、人権監視団は都市部の人口密集地域への強力な兵器の使用が増えているためだとしています。

一方、ウクライナのゼレンスキー大統領は14日、SNSに動画を投稿するとともに、ロシア中部のバシコルトスタン共和国や南部のクラスノダール地方の石油精製施設を攻撃したと発表しました。

また、黒海に面する街、ゲレンジークでは、ロシア産の石油を運ぶいわゆる「影の船団」のタンカーなどを攻撃したとしています。

ゼレンスキー大統領は「戦争を終わらせなければならない」としたうえで、ロシアへの圧力を強化し、和平に向けた交渉開始を迫りたい考えを示しました。

Translation

Ukraine's Civilian Casualties Exceed 2,200 in June: Highest in Four Years - UN

July 15, 2026, 8:33 AM

Ukraine Situation

On the 14th, the UN human rights monitoring team revealed that over 2,200 civilians were killed or injured in Ukraine last month due to Russia's ongoing military invasion, marking the highest monthly toll since April 2022 when the invasion began.

 The UN Observatory for Human Rights in Ukraine announced on the 14th that at least 2,283 Ukrainian civilians were killed or injured in the Russian military attacks last month.

The breakdown was 293 dead and 1,990 injured, the highest monthly figure since April 2022 when the invasion began.

The Human Rights Observatory attributed the increase in casualties to the increased use of powerful weapons against densely populated urban areas.

Meanwhile, on the 14th, Ukrainian President Zelenskyy posted a video on social media announcing attacks on oil refineries in the Republic of Bashkortostan in central Russia and the Krasnodar Krai in the south.

It was also stated that in the city of Gelendzhik which was facing the Black Sea, attacks were made on tankers carrying Russian oil, the so-called "shadow fleet".

President Zelenskyy declared that "the war must end" and indicated his intention to increase pressure on Russia and force the start of peace negotiations.

              So, the UN announces that at least 2,283 Ukrainian civilians were killed or injured in the Russian military attacks last month, the highest monthly figure since the invasion begins. Apparently, Ukraine is breeding and needs more air defense system to protect its people.

2026年7月28日 星期二

Apple Sues OpenAI: Alleged Theft of Trade Secrets

Recently NHK News on-line reported the following:

アップルがオープンAIを提訴企業秘密盗まれた”

202671112:53

アメリカ

アメリカのIT大手アップルは、自社の機密情報を盗んで独自の端末を開発しようとしているとして、生成AIChatGPTを開発するオープンAIを訴えました。

アップルが10日、西部カリフォルニア州の連邦地方裁判所に提訴したのは、アップルの元従業員で、現在、オープンAIで端末の開発に関わる2人と、オープンAIなどです。

アップルは訴状で「オープンAIは、技術スタッフからハードウエアの責任者に至るあらゆるレベルにおいて、アップルの企業秘密や機密情報を盗み取ってきた」として、組織的な不正行為だと主張しています。

具体的には、アップルからオープンAIへの転職を希望する従業員に対し、面接の際、製品のサンプルを持ってくるよう指示するなどして、製造技術を不正に取得したとしています。

オープンAIは去年、iPhoneなどのデザインを手がけたアップルの元幹部が率いるスタートアップ企業を買収し、新たにAIを搭載した端末の開発に乗り出しています。

オープンAIの広報担当者は「当社は、他社の企業秘密には一切、関心がない。世界中の人々に力を与える革新的な技術開発に注力する」とコメントしています。

両社はこれまでiPhoneに搭載するAIの機能で提携してきましたが、オープンAIが端末開発に乗り出したことを受けて、競合関係へと変化しています。

Translation

Apple Sues OpenAI: Alleged Theft of Trade Secrets

July 11, 2026, 12:53 PM

United States

American IT giant Apple sued OpenAI, the developer of the generative AI ChatGPT, alleging that OpenAI stole Apple's confidential information to develop its own devices.

On July 10, Apple filed the lawsuit in a federal district court in California against two former Apple employees currently involved in device development at OpenAI, as well as OpenAI itself.

In its complaint, Apple claimed that OpenAI had systematically stolen Apple's trade secrets and confidential information at every level, from technical staff to hardware managers.

Specifically, the company allegedly obtained manufacturing technology fraudulently by instructing employees who wanted to switch jobs from Apple to OpenAI to bring product samples to interviews.

Last year, OpenAI acquired a startup led by a former Apple executive who designed iPhones and other devices, and had since embarked on developing new AI-powered devices.

An OpenAI spokesperson commented, "We have no interest in other companies' trade secrets. We focus on developing innovative technologies that empower people around the world."

The two companies had previously collaborated on AI features for iPhones, but with OpenAI now venturing into device development, their relationship shifted to one of competition.

              So, American IT giant Apple sues OpenAI, alleging that OpenAI has stolen Apple's confidential information. Apparently, it is a common phenomenon that AI companies may be partners at one time and competitors at another time.

2026年7月26日 星期日

US-based OpenAI Begins Public Availability of New AI, Possibly Surpassing "Mythos"

Recently NHK News on-line reported the following:

米オープンAI 新型AI 一般への提供開始 「ミュトス」超えも

202671016:09

生成AI・人工知能

生成AIChatGPT」の開発を手がけるアメリカの「オープンAI」は、9日、最新のモデル「GPT5.6」の一般への提供を開始したと発表しました。

「オープンAI」は、9日、最新のAIモデル「GPT5.6」シリーズの一般への提供を始めたと発表しました。

これは、先月下旬以降、アメリカ政府の要請に応じて少数の企業などに限定して提供していたモデルで、今回対象を拡大した形です。

GPT5.6」は性能や価格ごとに3種類に分けられ、最上位のモデルでは、サイバーセキュリティーやプログラミングなどの性能を向上させたほか利用時のコストを抑えたのが特徴で、一部では、競合するAI新興企業「アンソロピック」が開発した「クロード・ミュトス5」を上回ったとしています。

「ミュトス5」はソフトウエアのぜい弱性を発見する能力が極めて高いとされるAIで、先月、アメリカ政府が安全保障上の懸念を示したことを受けて、会社は一時すべての顧客への提供を停止していました。

最先端のAIモデルをめぐっては、その性能の高さから高度なサイバー攻撃に悪用されるのではないかといった懸念が指摘されています。

Translation

US-based OpenAI Begins Public Availability of New AI, Possibly Surpassing "Mythos"

July 10, 2026, 16:09

Generative AI/Artificial Intelligence

OpenAI, the US company developing the generative AI "ChatGPT," announced on the 9th that it had begun publicly offering its latest model, "GPT5.6."

OpenAI announced on the 9th that it had begun publicly offering its latest AI model, the "GPT5.6" series.

This model had been offered to a limited number of companies since late last month at the request of the US government, and this time the targeted companies had been expanded.

“GPT5.6” was available in three versions according to performance and price. The top-end model features improved cybersecurity and programming capabilities while reducing operating costs. According to the company, in some performance areas, it surpassed Anthropic's “Claude Mythos 5.”

"Mythos 5" was an AI known for its extremely high ability to detect software vulnerabilities. Last month, the company temporarily suspended its provision to all customers after the U.S. government expressed national security concerns.

Concerns had been raised about this cutting-edge AI model due to its high performance, it could be misused for sophisticated cyberattacks.

              So, OpenAI announces that it has begun publicly offering its latest model, "GPT5.6." According to the company, in some performance areas, it surpasses Anthropic's “Claude Mythos 5.” Apparent, competition in making better AI model is heating up in the US.

2026年7月25日 星期六

[Analysis] What Does Trump's Announcement Regarding Interceptor Missiles for the "Patriot" System Mean for Ukraine?

 Recently CNN.co.jp reported the following:

【分析】「パトリオット」用迎撃ミサイルめぐるトランプ氏の発表、ウクライナにとって何を意味するのか

2026.07.09 Thu posted at 11:55 JST

CNN トランプ米大統領はウクライナに対し、防空システム「パトリオット」用の迎撃ミサイルの生産を許可したとみられる。これは二つの点で注目すべき動きだ。

まず、ウクライナはこれにより、米国が同盟国との共有を渋ってきた極めて機密性の高いミサイル防衛技術を利用できるようになる。数カ月前に米国から兵器供給の停止を言い渡されたウクライナだが、今や製造を許可される可能性が浮上しており、場合によっては、兵器を切実に必要とする米国や同盟国に一部を逆輸出する可能性すらある。

前途に待ち受けるハードルは多い。ウクライナは自前の生産ラインを構築しなければならないが、緊急の必要性や、ウクライナが何カ月も前からこの技術の複製を切望してきたことを考えると、スケジュールは大幅に短縮される可能性がある。

戦時下の国の動きは早い。ロッキード・マーチンやレイセオンといったパトリオットの製造メーカーは間違いなく、今回の取り決めについて疑問を持つだろう。これでは自社の持つ知的財産の商業的な将来性は、良く言っても薄められてしまう。ただ、ウクライナはドローン攻撃と迎撃技術の両面で世界をリードする実力を証明しており、新しいアイデアを素早く習得して大量生産を行い、現在の前線で優位に立ってきた。ウクライナがパトリオットを頻繁に使用することで、設計が改善されるかもしれない。

こうしたビジネス上の注目すべき動きから一歩引いて、トーンの大きな変化に注目してみよう。トランプ氏は昨年、ウクライナには切り札がないと主張して兵器の無償供与を停止し、自力購入を要求した。鉱物資源の譲渡を要求したり、情報共有を停止すると脅したりしたこともある。ところが今や、最も機密性の高いミサイルの製造を許可する意向を示し、ウクライナの攻勢でロシア政府に和平を強いることができる可能性に言及している。

これは驚くべき変化であり、クレムリン(ロシア大統領府)の耳にも届く可能性が高い。

Translation

[Analysis] What Does Trump's Announcement Regarding Interceptor Missiles for the "Patriot" System Mean for Ukraine?

July 9, 2026 (Thu) Posted at 11:55 JST

(CNN) US President Trump appeared to have authorized Ukraine to produce interceptor missiles for the "Patriot" air defense system. This would be a noteworthy development in two respects.

First, this would give Ukraine access to highly classified missile defense technology that the US had been reluctant to share with its allies. Several months ago, Ukraine, which was told by the US that arms supplies would be stopped, now had the possibility of being authorized to manufacture, and potentially even export some of the weapons back to the US and its allies that desperately needed them.

There were many hurdles waiting ahead. Ukraine had to build its own production lines, but given the urgent need and Ukraine's long-standing desire to replicate this technology, there was a possibility that the schedule might be significantly shortened.

Countries in wartime move quickly. Manufacturers of Patriot missiles, such as Lockheed Martin and Raytheon, would undoubtedly have doubts about this arrangement. At best, it diluted the commercial potential of their intellectual property. However, Ukraine had proven itself a world leader in both drone strike and interception technology, quickly adopting new ideas, mass-producing them, and gaining an advantage on the current front lines. Ukraine's frequent use of Patriots might even lead to design improvements.

Stepping back from these notable business developments, let's consider a significant shift in tone. Last year, Trump claimed Ukraine had no trump cards, halting free arms supplies and demanding they purchase them themselves. He also demanded the transfer of mineral resources and threatened to cut off intelligence sharing. Now, however, he's indicating his willingness to allow Ukraine to manufacture its most highly classified missiles and mentioning the possibility that a Ukrainian offensive could force the Russian government to make peace.

This was a surprising shift and was likely to reach the ears of the Kremlin (Presidential Administration of Russia).

So, there is a shift in tone in Trump. Months ago, he claimed that Ukraine had no trump cards, halting free arms supplies and demanding they purchase them themselves. Now, he's indicating a willingness to allow Ukraine to manufacture its most highly classified missiles and mentioning the possibility that a Ukrainian offensive could force the Russian government to make peace. It is interesting to note such changes.

2026年7月24日 星期五

阿里巴巴的人工智能大獲成功,但難以轉化為獲利模式(2/2)

Recently The New York Times reported the following:

Alibaba’s A.I. Is a Hit, but Hard to Turn into a Moneymaker (2/2)

The Chinese company’s models have won over developers worldwide, but they are open source — so they can be used and modified freely.

By Meaghan Tobin and Xinyun Wu - Reporting from Taipei, Taiwan

July 6, 2026, 12:05 p.m. ET

(continue)

In March, Lin Junyang, Qwen’s lead engineer, announced that he was leaving Alibaba. Several other key engineers left around the same time. Two people familiar with the team said it had become divided by disagreements over how best to commercialize Qwen.

Alibaba had long kept its largest, most advanced models proprietary, while releasing leading open-source models alongside them. Now, the company is signaling a broader shift away from widely used open-source models and toward closed ones that customers must pay to use. In April alone, Alibaba released three proprietary models within days of one another.

The Qwen team remains focused on keeping pace with the leading models coming out of Silicon Valley, but there is a growing recognition that technological leadership will not be enough if the company doesn’t make money, said a member of Alibaba’s A.I. research lab, which includes the Qwen team and others, who spoke on the condition of anonymity because the person was not authorized to discuss internal matters.

The challenge for Alibaba is apparent in its stock price. While A.I.-related stocks have surged in markets worldwide, Alibaba’s stock has fallen 37 percent this year in Hong Kong, where the broader market has declined 12 percent.

Building a top-performing A.I. model is expensive, requiring enormous investments in hangar-size data centers filled with computer chips that consume copious amounts of electricity.

Chinese A.I. companies face an additional hurdle: U.S. export controls limit their access to the most advanced chips. China’s leading A.I. start-ups and researchers routinely say shortages of computing power are the biggest thing holding them back, and they are spending heavily to secure it.

Those exorbitant costs are heaping financial pressure on Chinese A.I. companies. Start-ups like MiniMax and Z.ai have gone public to raise money from investors, and Alibaba is increasingly steering customers toward proprietary models even if that strategy upsets some of its top talent.

“It is a tough choice to make, and one that every open-model lab has to make at some point,” said Kevin Xu, the founder of Interconnected Capital, a hedge fund that invests in A.I. technologies.

Alibaba’s A.I. business faces mounting external challenges. The Pentagon recently placed the company on a blacklist of firms that it says support the Chinese military, a designation that Alibaba argues has already harmed its business. In a statement, Alibaba said it was not affiliated with the Chinese military.

Anthropic and OpenAI have also accused Chinese companies, including Alibaba, of improperly harvesting data from their A.I. systems to accelerate the development of their models.

Last month, Anthropic sent a letter to Senators Tim Scott, Republican of South Carolina, and Elizabeth Warren, Democrat of Massachusetts, accusing Alibaba of “brazenly” and “illicitly” trying to copy its technology using 24,000 fraudulent accounts. Alibaba declined to comment on the allegations.

Many in the technology industry compare today’s A.I. boom to the early days of the dot-com era, when internet companies were flourishing but no one had yet figured out how to build durable businesses around the technology.

Richard Lin, a vice president at the Silicon Valley company Datastrato who has long been involved in China’s open-source community, said China’s leading A.I. companies were all fighting for survival.

“There isn’t an A.I. company with a sustainable business model right now,” he said. “It’s not a healthy industry.”

Translation

阿里巴巴的人工智能大獲成功,但難以轉化為獲利模式(2/2

這家中國公司的模型贏得了全球開發者的青睞,但它們是開源的 - 因此可以自由使用和修改

(繼續)

今年3月,Qwen首席工程師 Lin Junyang 宣佈離開阿里巴巴。其他幾位關鍵工程師也在同一時間離職。兩位知情人士透露,團隊內部因如何更好地將 Qwen 商業化而產生分歧。

長期以來,阿里巴巴一直將其最大規模、最先進的模型設為專有模型,同時開放一些領先的開源模型。如今,該公司正發出信號,表明其正在從廣泛使用的開源模型轉向客戶必須付費使用的封閉模型。光是在4月份,阿里巴巴就在幾天之內發佈了三個專有模型。

阿里巴巴人工智能研究實驗室的一位成員表示,Qwen團隊仍然專注於去跟上矽谷的領先模式的步伐,但人們越來越意識到,如果公司無法獲利,僅僅擁有技術領先地位是不夠的。這位成員因未獲授權討論內部事務而要求匿名。

阿里巴巴面臨的挑戰在股價上顯而易見。儘管全球人工智能相關股票市場普遍上漲,但阿里巴巴的股價今年在香港已下跌37%,而香港整體股市也下跌了12%

建構一個性能卓越的人工智能模型成本高昂,需要巨額投資建設飛機庫般大的數據中心,這些數據中心堆滿了耗電量龐大的電腦晶片。

中國的人工智能公司也面臨另一個障礙:美國的出口管制限制了它們取得最先進晶片的管道。中國領先的人工智能初創公司和研究人員經常表示,運算能力短缺是限制他們發展的最大障礙,他們為此正在大量花錢來獲得它。

這些高昂的成本給中國的人工智能公司帶來了巨大的財務壓力。像MiniMaxZ.ai這樣的初創公司已經上市,從投資者那裡籌集資金,而阿里巴巴也越來越多地引導客戶使用其專有模型,即使這種策略會讓一些頂尖人才感到不滿。

投資人工智能技術的對沖基金 Interconnected Capital 的創始人 Kevin Xu 表示:「這是一個艱難的選擇,也是每個開放模型實驗室遲早都必須面對的選擇」。

阿里巴巴的人工智能業務面臨日益嚴峻的外部挑戰。五角大廈最近將該公司列入黑名單,稱其支持中國軍方,阿里巴巴認為這項認定已經對其業務造成了損害。阿里巴巴在聲明中表示,它與中國軍方沒有任何關聯。

Anthropic 和 OpenAI 也指責包括阿里巴巴在內的中國公司不正當地收集其人工智能資料以加速開發其模型系統。

上個月,Anthropic公司致函南卡羅來納州共和黨參議員Tim Scott和馬薩諸塞州民主黨參議員Elizabeth Warren,指控阿里巴巴“肆無忌憚”且“非法”地利用2.4萬個欺詐賬戶試圖複製其技術。阿里巴巴拒絕就此指控置評。

許多科技業人士將當今的人工智能熱潮比作網絡泡沫初期,當時網絡公司蓬勃發展,但沒有人找到圍繞這項技術,建立可持續的商業模式的方法。

長期參與中國開源界別的矽谷公司 Datastrato 副總裁 Richard Lin 表示,中國領先的人工智能公司都在為生存而戰。

他說:「目前沒有一家人工智能公司擁有可持續的商業模式」; 「這並非一個健康的行業」。

So, in China, Alibaba has re-emerged as an A.I. powerhouse, creating one of the world’s most widely used A.I. systems. They made its technology much cheaper to use than proprietary systems from U.S. competitors like Anthropic and OpenAI, helping the company attract users around the world. But open source A.I. models also raised a difficult question of how to turn that global popularity into business profit. Apparently, Alibaba needs to use earning from its other business such as e-commerce, cloud computing, and digital payments (through affiliated companies) etc. for support.

2026年7月22日 星期三

阿里巴巴的人工智能大獲成功,但難以轉化為獲利模式(1/2)

Recently The New York Times reported the following:

Alibaba’s A.I. Is a Hit, but Hard to Turn into a Moneymaker (1/2)

The Chinese company’s models have won over developers worldwide, but they are open source — so they can be used and modified freely.

By Meaghan Tobin and Xinyun Wu - Reporting from Taipei, Taiwan

July 6, 2026, 12:05 p.m. ET

Last year, the titans of China’s technology industry received a rare invitation to meet with Xi Jinping, the country’s leader, in the ceremonial Great Hall of the People.

But when the entrepreneurs — including the heads of national stars like the telecommunications giant Huawei and rising companies like the robotics start-up Unitree — assembled in neat rows before Mr. Xi, an unexpected figure was among them: Jack Ma.

Mr. Ma, who founded the Alibaba e-commerce empire in a cramped Hangzhou apartment and became a rock star of China’s internet era, had all but vanished from public view since Beijing torpedoed the planned $34 billion public offering of one of his companies after he criticized regulators in 2020.

Now, however, Alibaba has re-emerged as an A.I. powerhouse, creating one of the world’s most widely used A.I. systems.

Alibaba made its most popular A.I. models open source, allowing others to use and modify them freely. That made its technology much cheaper to use than proprietary systems from U.S. competitors like Anthropic and OpenAI, helping the company attract users around the world.

But it also raised a difficult question — how to turn that global popularity into a profitable business. It’s a challenge that many companies with open-source technologies face and one that is dividing Alibaba’s A.I. team.

Alibaba became one of China’s most valuable companies on the strength of its globe-spanning e-commerce business. But in recent years, it has been outmaneuvered by lower-cost rivals at home and has struggled to expand overseas. After Mr. Ma faded from public view, a new generation of executives took over, adopting a lower profile while emphasizing Alibaba’s role within Beijing’s broader technology policies.

“Alibaba is quite aware of its position as part of this team of national champions that are leading China’s charge in A.I.,” said Kyle Chan, a fellow at the Brookings Institution. “But also they have this huge responsibility to not repeat what they did before.”

In hindsight, Alibaba had gradually laid the groundwork to become an A.I. heavyweight. As its online shopping and logistics businesses expanded, the company built data centers capable of processing vast volumes of customer data worldwide.

It also borrowed a page from Amazon, parlaying its e-commerce success and technological infrastructure into a major cloud computing business, giving Alibaba two essential ingredients for building artificial intelligence systems: data and computing power.

“Alibaba in many ways has been ahead of its time,” Mr. Chan said. “It made early bets in a lot of these areas that are now the hot topics.”

Alibaba introduced its family of A.I. models, Qwen, in 2023 and quickly made them open source. When DeepSeek, an A.I. start-up, burst into the spotlight with claims that it had built a powerful model for a small fraction of the cost of Western competitors, global investors rushed to capitalize on the excitement around China’s open-source A.I.

China has made open-source technology a pillar of its drive to become an A.I. superpower. Nearly all of its leading A.I. systems are open source.

By January, Qwen had become the world’s most downloaded open-source A.I. system. Its models were being downloaded about one million times a day, according to data from Hugging Face, which hosts many open-source A.I. projects.

But that reach has not translated into big money. In the first three months of this year, Alibaba reported $1.3 billion in revenue from A.I.-related products — less than 4 percent of its total revenue. That pales in comparison with the company’s plan to spend more than $55 billion by the end of next year to build out its A.I. infrastructure.

The push to turn its open-source success into a profitable business has quietly fractured the team behind Qwen.

(to be continued)

Translation

阿里巴巴的人工智能大獲成功,但難以轉化為獲利模式(1/2

這家中國公司的模型贏得了全球開發者的青睞,但它們是開源的 - 因此可以自由使用和修改

去年,中國科技業的巨頭們獲得了一次難得的機會,在人民大會堂與國家領導人習近平會面。

但當這些企業家 - 包括國家之星華為等電信巨頭和Unitree等新興公司的負責人 - 整齊地列隊站在習近平主席面前時,一個意想不到的人物赫然出現在他們中間:馬雲。

馬雲先生在杭州一間狹小的公寓裡創立了阿里巴巴電商帝國,並成為中國網路時代的耀眼人物。然而,自從2020年他批評監管機構後,北京方面叫停了他旗下公司340億美元的IPO計劃,此後他幾乎從公眾視野中消失。

如今,阿里巴巴以人工智能巨頭的身份重返大眾視野,打造了全球應用最廣泛的人工智能系統之一。

阿里巴巴將其最受歡迎的人工智能模型開源,允許其他人自由使用和修改。這使得其技術的成本遠低於美國競爭對手(如AnthropicOpenAI)的專有系統,幫助公司吸引了全球用戶。

但這同時也引發了一個難題 - 如何將這種全球影響力轉化為獲利業務。這是許多擁有開源技術的公司面臨的挑戰,也是阿里巴巴人工智能團隊內部的分歧所在。

阿里巴巴憑藉著遍佈全球的電子商務業務,一躍成為中國市值最高的公司之一。但近年來,它在國內市場被成本較低的競爭對手超越,海外擴張也舉步維艱。馬雲淡出公眾視野後,新一代主管接手公司,行事低調,同時強調阿里巴巴在北京更廣泛的科技政策中扮演的角色。

布魯金斯學會研究員Kyle Chan表示:“阿里巴巴非常清楚自身是這支引領中國人工智能發展的國家冠軍團隊的一員。但他們亦肩負著巨大的責任而不能重蹈覆轍。”

回首往事,阿里巴巴其實早已為成為人工智能巨頭奠定了基礎。隨著其線上購物和物流業務的擴張,該公司在全球範圍內建立了能夠處理海量客戶資料的資料中心。

阿里巴巴也藉鑒了亞馬遜的成功經驗,將其電商成功和技術基礎設施轉化為強大的雲端運算業務,從而擁有建構人工智能系統的兩大關鍵要素:數據和運算能力。

Chan先生說:“阿里巴巴在很多方面都走在了時代的前端”,“它早期下注的許多前端領域如今已是熱門項目。”

阿里巴巴於2023年推出了其人工智能模型系列Qwen,並迅速將其開源。當人工智能初創公司DeepSeek聲稱其以遠低於西方競爭對手的成本構建了一個強大的模型時,全球投資者蜂擁而至,爭相從中國開源人工智能的熱潮中獲利。

中國已將開源技術作為其邁向人工智能超級大國的重要支柱。幾乎所有領先的人工智能系統都是開源的。

今年1月,Qwen已成為全球下載量最高的開源人工智能系統。根據託管眾多開源人工智能專案的 Hugging Face 資料顯示,其模型每天的下載量約為一百萬次。

但這種影響力並未轉化為豐厚的利潤。今年前三個月,阿里巴巴公佈的人工智能相關產品收入為 13 億美元,不到總收入的 4%。與該公司計劃在明年年底前投入超過 550 億美元用於建立人工智能基礎設施相比,這一數字顯得微不足道。

將開源的成功轉化為盈利業務的努力,悄悄導致了 Qwen 團隊的分裂。

(待續)

2026年7月21日 星期二

維基百科正為網絡的靈魂而戰(3/3)

Recently The New York Times reported the following:

Wikipedia Is Battling for the Soul of the Internet (3/3)

The internet’s largest stockpile of free knowledge is under threat from MAGA, A.I. and foreign autocrats. A bibliophile ex-ambassador is here to help.

The NYT - By Tiffany Hsu - Tiffany Hsu reports on the information ecosystem, including foreign influence, political speech and disinformation

July 5, 2026

(continue from part two)

A global challenge

Wikipedia sprawls across 345 different editions. The English one is by far the most active, with nearly 115,000 contributors, while others, in languages like the western African Kusaal and the Taiwanese Seediq, have smaller reach.

In countries like China, Myanmar and North Korea, Wikipedia is blocked entirely. Turkey banned the site for nearly three years until 2020. Wikimedia negotiated with the Indonesian government this year over blackout threats.

Since 2020, at least 10 Wikipedia editors have been imprisoned for their work and countless others threatened. A pediatrician in Saudi Arabia was sentenced to 14 years in prison after editing pages about government surveillance and a women’s rights activist. Belarus has detained several editors, sending one to a penal colony for more than a year after he edited a post about murdered journalists. Curbs in free expression have been rising globally, affecting Wikipedia and its volunteers, a foundation spokesman wrote in an email.

Governments are also targeting Wikipedia content specifically. Of the 934 requests the foundation received last year to take down or alter content (it granted two), 117 came from governments, up from 10 a decade ago.

Most were from Russia, whose antagonism toward Wikipedia deepened after the country invaded Ukraine in 2022. Moscow has repeatedly accused the site of violating a sweeping censorship law that criminalized content deemed to “discredit” Russia’s military (Wikimedia refuses to pay the associated fines). A Russian clone of Wikipedia called Ruwiki was introduced in 2023; a Russian network of fake news portals managed to infiltrate some Wikipedia citations.

Wikimedia is expanding its human rights team, a division of the legal department that safeguards volunteers from digital and physical threats. The foundation recently deployed an anti-doxxing initiative offering contributors more anonymity; it has also connected Wikipedians to local evacuation and extraction services.

Defending volunteers and projects consumes 32 percent of the foundation’s budget. Wikimedia’s litigation costs have doubled since 2020, in part because there are more cases, which new privacy and speech regulations are making more complex.

“A core component of what we do is preserving freedom to information,” Ms. Meehan said. “That will never change, that commitment to advocating for those rights, and to helping to protect within our limited means editors who engage in that project and are persecuted for that very reason.”

A search for new recruits

Wikipedia’s future has often seemed precarious — its entry titled “Predictions of the end of Wikipedia” is nearly nine years old. But especially now, with so many dangers at its gate and fewer people joining its ranks, the site cannot afford to grow stale.

The foundation wants to entice a new generation of contributors that is not predominantly white, male and graying. Training sessions to demystify editing, for example, drew one 25-year-old Vietnamese American volunteer who has created hundreds of pages in two years about zeitgeisty topics like the “swag gap” and the “performative male.”

In five months, Ms. Meehan has met more than 1,100 such Wikipedians across four continents. Unlike most modern content creators, they “aren’t out there looking to be the next influencer or famous person,” she said. When they congregate, as on a recent rainy weekend in San Francisco, they want to discuss the finer points of A.I. detection and disappearing databases.

Wikipedians aren’t flashy, but they are tough. Two of them tackled a gunman at a conference last fall. Ms. Meehan fits in, always wearing a simple black Ironman watch, after the fancy Baume & Mercier timepiece she purchased when she first joined the foreign service was stolen 20 years ago in Colombia by armed assailants.

Seeing the depravity of war and the worst of humanity fundamentally changed her, she said, compelling her to “run toward the light” and seek out “the antithesis of the bad.” That’s Wikipedia, she said — an experiment that will keep fighting to push society toward conciliation over conflict.

“In the world right now, where politicization is on the rise, where there’s disagreement on facts, where people can’t even engage civilly in a conversation, it’s OK to disagree,” she said. “But let’s come together and have a conversation based on fact and see where that takes us.”

Translation

維基百科正為網絡的靈魂而戰(3/3

網絡上最大的免費知識寶庫正受到「讓美國再次偉大」(MAGA)、人工智能和外國獨裁者的威脅。一位熱愛書籍的前大使將伸出援手

 (接續第二部分)

全球挑戰

維基百科擁有345個不同的版本。英文版最活躍,有近11.5萬名獻稿者,而其他語言版本,例如西非庫薩爾語和台灣賽德克語,有較小獻稿。

在中國、緬甸和北韓等國家,維基百科被完全封鎖。土耳其曾封鎖該網站近三年,直到2020年才解封。今年,維基媒體與印尼政府就封鎖威脅進行了談判。

2020年以來,至少有10名維基百科編輯因其工作而被捕入獄,另有無數人受到威脅。沙烏地阿拉伯一名兒科醫生因編輯有關政府監控和一名女權運動家的頁面而被判處14年監禁。白俄羅斯拘留了多名編輯,其中一人因編輯一篇關於遇害記者的文章而被送往勞改營一年多。維基百科基金會發言人在一封電子郵件中寫道,全球對言論自由的限制日益加劇,這影響了維基百科及其義工。

各國政府也專門針對維基百科的內容。去年,維基百科基金會收到了934份刪除或修改內容的請求(其中2份獲得批准),其中117份來自政府,而十年前這一數字僅為10份。

大多數是來自俄羅斯,俄羅斯對維基百科的敵意在2022年入侵烏克蘭後加劇。莫斯科多次指責維基百科違反了一項廣泛的審查法,該法將被認為「詆毀」俄羅斯軍方的內容定為犯罪(維基媒體拒絕支付相關罰款)。一個名為Ruwiki的維基百科俄羅斯複製網站於2023年出現;一個俄羅斯假新聞網站網絡成功滲透了一些維基百科的引用。

維基媒體正在擴充其人權團隊,該團隊隸屬於法律部門,負責保護義工免受網絡和人身威脅。該基金會最近推出了一項反人肉搜尋倡議,為貢獻者提供更高的匿名性;它還為維基百科的人去聯繫當地的疏散和撤離服務。

保護義工及其計劃消耗了基金會32%的預算。自2020年以來,維基媒體的訴訟成本翻了一番,部分原因是案件數量增加,而新的隱私和言論法規使案件變得更加複雜。

Meehan女士說: 「我們工作的核心在於維護資訊自由」; 「我們永遠不會改變這一承諾,我們將致力於倡導這些權利,並在能力所及的範圍內幫助那些參與維基百科項目並因此遭受迫害的編輯們」。

招募新成員

維基百科的未來似乎一直岌岌可危 - 其題為「關於維基百科終結的預測」的條目至今已有近九年歷史。但尤其是在如今,面臨諸多挑戰且在新成員不斷減少的情況下,維基百科更不能停滯不前。

維基百科基金會希望吸引新一代的獻稿者,不再以白人、男性和老年人為主。例如,為了普及編輯知識,維基百科舉辦了培訓課程,吸引了一位25歲的越南裔美國義工。這位義工在兩年內創建了數百個頁面,內容涉及「時尚差距」和「表演型男性」等潮流話題。

五個月來,米漢Meehan女士在四大洲認識了超過1,100位維基百科人。她說,與大多數現代內容創作者不同,他們「並非一心想成為下一個網紅或名人」。當他們聚在一起時,例如最近在舊金山一個下雨的能週末,他們更想探討人工智能偵測和資料中心消失等更細節問題。

維基百科人並不張揚,但他們堅韌不屈。去年秋天,其中有兩位在一次會議上制服了一名持槍歹徒。Meehan女士符合這種特點,她總是戴著一隻簡約的黑色Ironman手錶。 20年前,她剛加入外交部門時購買的名士(Baume & Mercier)華麗名表在哥倫比亞被武裝份子搶走。

她說,目睹戰爭的殘酷和人性醜惡徹底改變了她,促使她“奔向光明”,去尋找“邪惡的對立面”。她說,這就是維基百科 - 一個會繼續努力推動社會走向和解而非衝突的實驗。

她說:「在當今世界,政治化日益加劇,人們對事實存在分歧,甚至無法進行文明對話,在這種情況下,意見不一致是可以接受的」; 「但讓我們走到一起,基於事實展開對話,看看這樣做會把我們帶到哪裡」。

              So, Wikipedia is in a difficult situation. The MAGA right, with Elon Musk at the fore, is questioning the work of this web site. Repressive governments have hauled Wikipedia’s volunteer editors into penal colonies. Wikipedia is globally one of the most important information centers and companies like Google and Meta have devoured the encyclopedia’s 65 million articles, and the site is controlling the access of certain scrapers every day to avoid over burdening its system. As an organization that holds neutrality as a cardinal rule, it faces conflicts, and enemies. May I wish Bernadette Meehan every success in her new post.

Note:

1.  About the term "Swag gap"(時尚差距) ,  swag means someone's style, confidence, charisma, or "cool factor." It comes from African American Vernacular English (AAVE) and became mainstream in the 2000s. The swag gap is the perceived difference in coolness, charisma, or social appeal between two people or groups. (ChatGPT)

2. About the term "Performative male" (表演型男性), the adjective performative means doing something mainly to create an impression rather than because it is genuine. So, a performative male is a man whose behavior or masculinity seems designed for an audience rather than being authentic. ChatGPT)

2026年7月20日 星期一

維基百科正為網絡的靈魂而戰(2/3)

Recently The New York Times reported the following:

Wikipedia Is Battling for the Soul of the Internet (2/3)

The internet’s largest stockpile of free knowledge is under threat from MAGA, A.I. and foreign autocrats. A bibliophile ex-ambassador is here to help.

The NYT - By Tiffany Hsu - Tiffany Hsu reports on the information ecosystem, including foreign influence, political speech and disinformation

July 5, 2026

(continue from part one)

A fight with the right

Mr. Musk has railed against Wikipedia as “Wokepedia” and “an extension of legacy media propaganda.” He has protested his entry and urged his followers to withhold donations over the foundation’s diversity initiatives.

The tech mogul is one in a chorus of conservatives calling the reference site a hotbed of liberal bias. Senator Ted Cruz, Republican of Texas, expressed “alarm” about both Wikipedia’s purported ideological lean and claims that editors were coordinating to spread antisemitic and pro-Hamas narratives. David Sacks, recently the Trump administration’s A.I. czar, weighed in. So did Tucker Carlson, a popular commentator, and even Larry Sanger, who founded Wikipedia with Jimmy Wales in 2001.

Wikipedia has long been a proxy battleground for warring ideas about truth: What constitutes balance? What makes a good source? But now, those online debates could result in real repercussions.

Last year, a Trump administration official made inquiries into the nonprofit’s tax-exempt status. A Republican-led congressional oversight committee initiated an investigation in August, demanding that Wikimedia identify certain editors and produce any evidence of influence from foreign operatives or academic institutions. The foundation said it responded to both inquests last year and had no updates.

To de-escalate the antagonism, Ms. Meehan and her team are addressing misconceptions about the site. Wikipedia is largely decentralized, with independent contributors distilling facts through open debate and agreement. Elected arbitration committees handle dispute resolution. The foundation, with a budget of $208.6 million and about 600 employees and contractors, provides servers and funding but steers clear of content decisions.

There is no shadowy mastermind, no leftist cabal, according to the foundation. Just the Wikipedians — nearly 250,000 unpaid strangers editing 324 times a minute under pseudonyms like “Dr. Blofeld” and “WooHooKitty.”

“The most important thing to understand about Wikipedia is it is truly a democracy,” said Bill Adair, who created the fact-checking site PolitiFact and is writing a book about the encyclopedia. “It’s messy, it’s loud, it often works.”

Ms. Meehan has been hauling her message to Capitol Hill, where she’s distributed personally meaningful printouts of Wikipedia entries. The foundation also uses lobbyists, nearly doubling its spending to $174,300 in the most recent fiscal year and retaining the Holland & Knight firm last spring for $150,000 a quarter.

“We are not seeking to be part of culture wars, of political discourse,” Ms. Meehan said. “These are people who want to put knowledge into the world, for people to do with it what they want.”

To those convinced of bias on Wikipedia, her recommendation is simple: Become an editor.

“More dialogue, frankly, is what’s needed in this moment,” she said. “So the door is open.”

Gorging on a data buffet

For many Wikipedians, the Trump administration may be the site’s most acute threat, but its most existential one is artificial intelligence. They have resisted the idea of integrating the technology, calling it “truly ghastly” (or, simply, “yuck”).

The more urgent worry, though, is that Wikipedia is being exploited to train its own competitors, as A.I. systems hoover up its content to inform chatbots like Gemini, Claude and ChatGPT. The bots then regurgitate the information, often imperfectly, polluting the information ecosystem that feeds into the encyclopedia.

Exhibit A: Grokipedia, a rival built by Mr. Musk and edited by xAI, his artificial intelligence company. It has no public edit history and lifts liberally from Wikipedia while also inventing errors (including several in its profile of Ms. Meehan).

A.I.-generated summaries atop search results also siphon away potential visitors, she said. Human page views of Wikipedia’s English edition slumped 8 percent late last year compared with a year earlier. The decline is concerning enough that Ms. Meehan — whose favorite entry describes “tsundoku,” the Japanese term for collecting books and letting them pile up unread — is pushing to reach younger people through TikTok and Roblox.

Wikimedia wants more give-and-take from the A.I. industry. It now charges clients like Amazon and Microsoft for faster bulk access to its data through a four-year-old commercial subsidiary called Enterprise. Last year, Enterprise earned $8.3 million, more than doubling the prior year’s revenue.

This spring, Wikimedia also began limiting automated freeloaders from scraping its site and overwhelming its back end — which accounts for nearly a third of its most intensive bandwidth demands. But the foundation can currently halt only about 30 percent of abusive data requests — about 1.5 billion daily.

“Our infrastructure is not free, and when scrapers come in and bulk-download, it really takes a toll,” Ms. Meehan said. “There is a literal dollar cost to that behavior.”

(to be continued by part three)

Translation

維基百科正為網絡的靈魂而戰(2/3

網絡上最大的免費知識寶庫正受到「讓美國再次偉大」(MAGA)、人工智能和外國獨裁者的威脅。一位熱愛書籍的前大使將出手相助

(接上文)

與右翼的鬥爭

Musk先生猛烈抨擊維基百科,稱其為「覺醒百科」(Wokepedia)和「傳統媒體宣傳的延伸」。他抗議自己被收錄到維基百科,並敦促其追隨者因維基百科基金會的多元化舉措而停止捐款。

這位科技巨頭是許多保守派人士中的一員,他們稱維基百科是自由派偏見的溫床。德州共和黨參議員Ted Cruz對維基百科所謂的意識形態傾向以及編輯們串通傳播反猶太主義和親哈馬斯言論的說法表示「震驚」。近期的特朗普政府人工智能主專員David Sacks也發表了看法。知名評論員Tucker Carlson,甚至維基百科的創始人之一Larry Sanger(他與Jimmy Wales2001年共同創立了維基百科)也加入了爭論。

長期以來,維基百科一直是關於什麼是真相的各種爭論的代理人戰場:什麼是平衡?什麼是是可靠的資訊來源?如今,這些網絡辯論可能會產生實際影響。

去年,一位特朗普政府官員對維基百科的免稅資格進行了調查。一個由共和黨領導的國會監督委員會於8月啟動了一項調查,要求維基媒體基金會指定一些編輯,去提供有或没有外國勢力或學術機構施加影響的證據。該基金會表示,他們已在去年對這兩項調查作出回應,目前沒有新的進展。

為了緩和對抗,Meehan女士和她的團隊正在努力消除人們對維基百科的誤解。維基百科在很大程度上是去中心化的,獨立的獻稿者透過公開辯論和協商來提取事實。由選舉產生的仲裁委員會負責處理爭議。維基百科基金會擁有2.086億美元的預算和約600名員工及承包商,負責提供伺服器和資金,但不參與內容決策。

據基金會稱,維基百科背後沒有幕後主使,也沒有左翼陰謀集團。維基百科完全由維基百科人主導 - 25萬名素不相識的無償獻稿者所運作,他們使用「Dr. Blofeld」和「WooHooKitty」等化名,每分鐘編輯324次。

創建了叫PolitiFact的事實核查網站,並正在撰寫一本關於維基百科書籍的Bill Adair說道:「理解維基百科最重要的一點是,它確實是一個民主的平台」,「它或許有些混亂,或許有些喧鬧,但它往往行之有效」。

Meehan女士一直在國會山莊宣傳她的理念,她在那裡分發了對她個人意義非凡的維基百科項目的打印件。該基金會還聘請了遊說人員,在最近一個財政年度,其支出幾乎翻了一番,達到174,300美元,並在去年春天以每季150,000美元的價格聘請了Holland & Knight 律師事務所。

Meehan女士說: 「我們無意捲入政治話語的文化戰爭」; 「大家只是想把知識傳播到世界各地,讓人們可以自由地運用它」。

對於那些認為維基百科有偏見的人,她的建議很簡單:成為一名維基百科編輯。

她說:「坦白說,此刻我們需要的是更多的對話」,「所以,大門永遠是敞開」。

數據盛宴

對許多維基百科用戶來說,特朗普政府或許是該網站面臨的最嚴峻威脅,但對其生存而言,最根本的威脅卻是人工智能。他們一直在抵制與人工智能協調整合,稱其「實在可怕」(或乾脆說「噁心」)。

然而,更令人擔憂的是,維基百科正被利用來訓練競爭對手,因為人工智能系統會大量收集其內容,用於訓練諸如 GeminiClaude ChatGPT 之類的聊天機器人。這些機器人隨後會重述這些訊息,而且往往不夠完善,因此污染了維基百科的資訊生態系統。

例證一:Grokipedia,這是Musk先生創建並由其人工智能公司 xAI 編輯的競爭對手。它沒有公開的編輯歷史,大量抄襲維基百科的內容,同時也捏造錯誤(包括其對Meehan女士的個人資料中的幾處錯誤)。

她還表示,AI 生成的摘要在搜尋結果的頂部,也吸走了潛在的訪客。去年年底,維基百科英文版的頁面的真人瀏覽量比去年同期下降了 8%。這種下滑趨勢令人擔憂,令Meehan女士(她最喜歡的詞條是去描述“tsudoku”,即日語中指囤積書籍卻不讀的現象)要努力地去透過TikTokRoblox等平台吸引年輕用戶。

維基媒體希望與人工智能產業建立更多互惠互利的關係。現在它透過一家成立已有四年名為Enterprise的商業子公司向亞馬遜和微軟等客戶收取費用,獲得可以加快大量存取維基百科資料的速度。 Enterprise的收入為830萬美元,比前一年翻了一番還多。

今年春天,維基媒體也開始限制外間的自動程式去抓取其網站資料,因為它們給伺服器和其他維持網站運作的系統帶來了過大的壓力 - 這幾乎佔其最密集的頻寬需求的三分之一。但維基目前只能阻止約30%的濫用數據請求 - 每天約15億次。

Meehan女士說: 「我們的基礎設施並非免費,當網絡抓取機器人入侵並進行批量下載時,會造成巨大的損失」; 「這種行為會造成實實在在的經濟損失」。

(待續,見第三部分)