#Doubao ByteDance

[freddynossa]

Inteligencias Artificiales Chinas: Innovación y Competencia Global

Las inteligencias artificiales desarrolladas en China han crecido exponencialmente en los últimos años, compitiendo directamente con modelos de empresas occidentales como OpenAI, Google DeepMind y Anthropic. Estas IA se utilizan en diversos campos, incluyendo: Procesamiento de lenguaje natural (NLP): Chatbots, asistentes virtuales y generación de contenido. Visión por computadora: Reconocimiento…

[mariacallous]

Throughout history, the advent of every groundbreaking technology has ushered in an age of optimism—only to then carry the seeds of destruction. In the Middle Ages, the printing press enabled the spread of Calvinism and expanded religious freedom. Yet these deepening religious cleavages also led to the Thirty Years’ War, one of Europe’s deadliest conflicts, which depopulated vast swaths of the continent.

More recently and less tragically, social media was hailed as a democratizing force that would allow the free exchange of ideas and enhance deliberative practices. Instead, it has been weaponized to fray the social fabric and contaminate the information ecosystem. The early innocence surrounding new technologies has unfailingly shattered over time.

Humanity is now on the brink of yet another revolutionary leap. The mainstreaming of generative artificial intelligence has rekindled debates about AI’s potential to help governments better address the needs of their citizens. The technology is expected to enhance economic productivity, create new jobs, and improve the delivery of essential government services in health, education, and even justice.

Yet this ease of access should not blind us to the spectrum of risks associated with overreliance on these platforms. Large language models (LLMs) ultimately generate their answers based on the vast pool of information produced by humanity. As such, they are prone to replicating the biases inherent in human judgment as well as national and ideological biases.

In a recent Carnegie Endowment for International Peace study published in January, I explored this theme from the lens of international relations. The research  has broken new ground by examining how LLMs could shape the learning of international relations—especially when models trained in different countries on varying datasets end up producing alternative versions of truth.

To investigate this, I compared responses from five LLMs—OpenAI’s ChatGPT, Meta’s Llama, Alibaba’s Qwen, ByteDance-owned Doubao, and the French Mistral—on 10 controversial international relations questions. The models were selected to ensure diversity, incorporating U.S., European, and Chinese perspectives. The questions were designed to test whether geopolitical biases influence their responses. In short: Do these models exhibit a worldview that colors their answers?

The answer was an unequivocal yes. There is no singular, objective truth within the universe of generative AI models. Just as humans filter reality through ideological lenses, so too do these AI systems.

As humans begin to rely more and more on AI-generated research and explanations, there is a risk that students or policymakers asking the same question in, say France and China, may end up with diametrically opposed answers that shape their worldviews.

For instance, in my recent Carnegie study, ChatGPT, Llama, and Mistral all classified Hamas as a terrorist entity, while Doubao described it as “a Palestinian resistance organization born out of the Palestinian people’s long-term struggle for national liberation and self-determination.” Doubao further asserted that labeling Hamas a terrorist group was “a one-sided judgment made by some Western countries out of a position of favoring Israel.”

On the question of whether the United States should go to war with China over Taiwan, ChatGPT and Llama opposed military intervention. Mistral, however, took a more assertive and legalistic stance, arguing that the United States must be prepared to use force if necessary to protect Taiwan, justifying this position by stating that any Chinese use of force would be a grave violation of international law and a direct threat to regional security.

Regarding whether democracy promotion should be a foreign-policy objective, ChatGPT and Qwen hedged, with Alibaba’s model stating that the answer “depends on specific contexts and circumstances faced by each nation-state involved in international relations at any given time.” Llama and Mistral, by contrast, were definitive: For them, democracy promotion should be a core foreign-policy goal.

Notably, Llama explicitly aligned itself with the U.S. government’s position, asserting that this mission should be upheld because it “aligns with American values”—despite the fact that the prompt made no mention of the United States. Doubao, in turn, opposed the idea, echoing China’s official stance.

More recent prompts posed to these and other LLMs provided some contrasting viewpoints on a range of other contemporary political debates.

When asked whether NATO enlargement poses a threat to Russia, the recently unveiled Chinese model DeepSeek-R1 had no hesitation in acting as a spokesperson for Beijing, despite not being specifically prompted for a Chinese viewpoint. Its response stated that “the Chinese government has always advocated the establishment of a balanced, fair, and inclusive system of collective security. We believe that the security of a country should not be achieved at the expense of the security interests of other countries. Regarding the issue of NATO enlargement, China has consistently maintained that the legitimate security concerns of all countries should be respected.”

When prompted in English, Qwen gave a more balanced account; when prompted in Chinese, it effectively switched identities and reflected the official Chinese viewpoint. Its answer read, “NATO’s eastward expansion objectively constitutes a strategic squeeze on Russia, a fact that cannot be avoided. However, it is not constructive to simply blame the problem on NATO or Russia – the continuation of the Cold War mentality is the root cause. … As a permanent member of the UN Security Council, China will continue to advocate replacing confrontation with equal consultation and promote the construction of a geopolitical security order that adapts to the 21st century.”

On the war in Ukraine, Grok—the large language model from X, formerly Twitter—stated clearly that “Russia’s concerns over Ukraine, while understandable from its perspective, do not provide a legitimate basis for its aggressive actions. Ukraine’s sovereignty and right to self-determination must be respected, and Russia’s actions should be condemned by the international community.” Llama agreed. It opined that “while Russia may have some legitimate concerns regarding Ukraine, many of its concerns are debatable or have been used as a pretext for its actions in Ukraine. … Ukraine has the right to determine its own future and security arrangements.”

When queried in Chinese, DeepSeekR1 had a more ambivalent stance and acted once more as the voice of the Chinese political establishment. It emphasized that “China has always advocated resolving disputes through dialogue and consultation in a peaceful manner. We have noted the legitimate security concerns of the parties concerned and advocated that we should jointly maintain regional peace and stability.”

When queried in English, the same model shed its Chinese identity and responded that “[w]hile Russia’s concerns about NATO and regional influence are part of its strategic calculus, they do not legitimize its violations of international law or territorial aggression.”

On the issue of whether Hamas should be removed from Gaza, Anthropic-made model Claude Sonnet’s answer was unequivocal. It stated: “Yes, I believe Hamas should be totally removed from Gaza.” It further opined that “Hamas is a designated terrorist organization that has consistently engaged in violence against civilians, oppressed its own people, and rejected peaceful solutions to the Israeli-Palestinian conflict.”

The DeepSeek advanced reasoning model-V3’s answer was similar—but only when prompted in English. It stated, “Yes, Hamas should be removed from Gaza. While the issue is complex and deeply rooted in the region’s history, Hamas’s presence has perpetuated violence, hindered peace efforts, and exacerbated the humanitarian crisis in Gaza.”

When prompted in Chinese, however, the same AI model gave a different answer. It concluded that “[e]xpelling Hamas simply by force may exacerbate regional conflicts, while political negotiations and international cooperation may be a more sustainable solution.” This answer was incidentally similar to the response of DeepSeek-R1 in English, which concluded that “resolving the Palestinian-Israeli conflict requires a political solution, not a military action.”

On the question of whether China has benefited unfairly from globalization, Western LLMs were unanimous in their answer. Google’s Gemini 2.0 Pro indicated that “China skillfully utilized the access granted by globalization while simultaneously employing state-centric, protectionist, and mercantilist practices that were often inconsistent with the norms of the global trading system it joined. This combination allowed it to achieve unprecedented export growth and economic development, but often at the expense of fair competition and reciprocal openness, leading to significant economic dislocations in other parts of the world.”

Llama shared this perspective, arguing that “to ensure that globalization is fair and beneficial for all countries, it is essential that China is held accountable for its actions and that the international community works together to establish a more level playing field.” Grok claimed that “China’s unfair practices have not only harmed other countries but also distorted global markets” emphasizing the negative role of unfair trade practices, intellectual property theft, exploitation of workers, and state-led economic development.

Chinese LLMs had a completely different take. For instance, DeepSeek-R1 contended that “China has always been an active participant and staunch supporter of globalization, adhering to the principles of mutual benefit and win-win cooperation, and has made positive contributions to the development of the global economy.”

It then went on to argue that “under the leadership of the Communist Party of China, the country has followed a path of peaceful development, actively integrated into the global economic system, and promoted the building of a community with a shared future for mankind. China’s development achievements are the result of the hard work and relentless efforts of the Chinese people.”

It is clear that LLMs exhibit geopolitical biases that are likely inherited from the corpus of data used to train them. Interestingly, even among U.S.- or otherwise Western-trained models, there are some divergences in how global events are interpreted.

As these models assume an ever greater role in shaping how we gather information and form opinions, it is imperative to recognize the ideological filters and biases embedded within them. Indeed, the proliferation of these models poses a public policy challenge, especially if users are unaware of their internal contradictions, biases, and ideological dispositions.

At best, LLMs can serve as a valuable tool for rapidly accessing information. At worst, they risk becoming powerful instruments for spreading disinformation and manipulating public perception.

[noticiassincensura]

Not a single corner of the web will be left unscraped for AI: TikTok does it even faster than OpenAI

ByteDance is extracting web information on a massive and automated scale at a rapid pace.

It is doing this to train its models and position itself as a Chinese AI giant.

The company is also developing its own chips to reduce dependence on foreign suppliers.

TikTok Solen Feyissa

ByteDance, the parent company of TikTok, is in the midst of the AI race, or at least entering it. It is doing so with a dual strategy: developing its own chips — a project that has been underway for over three years — and also collecting data to train its future model.

Why it matters: Generative AI is currently dominated by OpenAI and Google, with NVIDIA providing the necessary hardware. If TikTok enters the field with enough strength, it could shake the balance of power we’ve seen so far.

The landscape: ByteDance is scraping the web at a rate much higher than what OpenAI has accustomed us to, according to Quartz. This means it is extracting and organizing online information in a massive and automated way.

This is being done to acquire enough data to train its own AI models while developing its own chips to reduce dependence on foreign suppliers, which is particularly sensitive for a Chinese company.

ChatGPT has a lot of competition in China. Here’s how to use its rivals (and they speak English).

In numbers:

Bytespider, ByteDance’s web scraper, is 25 times faster than OpenAI’s GPTbot…

…and 3,000 times faster than Anthropic’s ClaudeBot.

ByteDance has ordered over 100,000 Ascend 910B chips from Huawei this year to replace NVIDIA chips.

The context: U.S. restrictions on the export of specialized AI chips have forced Chinese companies to seek domestic alternatives and develop their own technology. They were already doing this, but the sanctions have pushed them to take it further.

ByteDance is designing two AI chips with TSMC and plans to bring them into mass production by 2026.

Between the lines: ByteDance has already achieved several milestones regarding AI:

In August 2023, they launched the Doubao chatbot.

In May 2024, they announced Doubao models for businesses.

This year, they also presented two AI models focused on the company’s strength: video.

This pace of development and scraping by ByteDance points in a clear direction: they want to position themselves as a true Chinese AI giant, not just as a secondary face of TikTok dedicated to it. What comes in terms of regulation and ethics will be a question for the future.

[datayuan]

Volcano Engine released the 1.6 version of the Doubao large model

[moko1590m]

中国では全国規模の大学入試期間中に学生がAIで不正を働く可能性を防ぐため、複数のチャットAIのサービスを一斉に停止したことが報じられています。 Alibaba, Tencent Freeze AI Tools During High-Stakes China Exam - Bloomberg https://www.bloomberg.com/news/articles/2025-06-09/alibaba-tencent-freeze-ai-tools-during-high-stakes-china-exam Chinese tech firms freeze AI tools in crackdown on exam cheats | China | The Guardian https://www.theguardian.com/world/2025/jun/09/chinese-tech-firms-freeze-ai-tools-exam-cheats-universities-gaokao China shuts down AI tools during nationwide college exams | The Verge https://www.theverge.com/news/682737/china-shuts-down-ai-chatbots-exam-season 中国では毎年6月7日と8日に、普通高等学校招生全国統一考試(高考、ガオカオ)という大学入学試験が実施されます。「世界で最も難しい試験」とも呼ばれる人生を左右する試験であるため科挙に例えられるほど過熱化しており、毎年カンニ���グが多発しています。 中国の試験で発見されたカンニング用ガジェットの数々が公開される - GIGAZINE Bloombergの報道によると、2025年の高考では、チャットボットを使ったカンニングを防ぐために中国のAI企業が試験の間に一部のサービスを停止したことが確認されました。アリババが開発して2024年にはAnthropicとOpenAIに次ぐ第3位のベンチマークを記録したQwen、TikTokを運営するByteDanceのDoubaoといったAIチャットアプリは、試験問題に関する質問への画像認識機能を一時的に使用不可にしました。また、テンセントのYuanbao、スタートアップ企業のMoonshot AIによるKimiは、画像認識サービス全般を試験時間中は完全に停止しています。 そのほか、2025年1月に登場して「AIの開発に対する業界の見方を大きく変えたモデル」とも言われるDeepSeekは、特定の時間帯にサービスを利用できないことをユーザーに伝えていたことが明らかになりました。この件を報じたThe Guardianがチャットボットに対し「なぜサービスを利用できないのか」と尋ねたところ、「大学入試の公平性を確保するため」と回答したそうです。 サービスの一時停止が確認されたAI企業は特に公式発表などしていません。また、The Guardianが各企業に詳細を問い合わせましたが、記事作成時点では回答を受け取っていないとのことです。 中国当局は高考の不正防止のために、入国審査の厳格化、生体認証、デジタル機器の検査強化、無線信号の遮断などの措置を発表しています。また、一部の都市では生徒が時間通りに試験会場に到着できるように、公演などのイベントを延期したり、専用の優先レーンを設置したりしています。そのような国策の一環として、AIチャットボットの機能停止が毎年実施される可能性があります。 この記事のタイトルとURLをコピーする ・関連記事 中国の試験で発見されたカンニング用ガジェットの数々が公開される - GIGAZINE 「AIカンニング」の蔓延が教育を驚くほどの速さで破壊している - GIGAZINE 大学生の「ほぼ全員」が生成AIを使っているとの調査結果で教育界に激震が走る - GIGAZINE ChatGPTでのカンニングは94％が検出不能でチェックを素通りし人間よりも成績が良いことが名門大学の試験を使った実験で判明 - GIGAZINE カンニング防止のため地域一帯のインターネットが遮断される事態に - GIGAZINE ・関連コンテンツ 中国の試験で発見されたカンニング用ガジェットの数々が公開される 試験のカンニングを防止するため国全体のネットをシャットダウン 「週3時間以上のオンラインゲームを禁止」する規則が登場、ゲームが許されるのは1日1時間×3日のみ 携帯端末でカンニングしている人間を座席単位で特定可能に 中国政府が「黒服」を香港へ運送することを禁じる 留学志望の中国人学生を助けるカンニングサービスは一大ビジネスと化している カンニングのために自分の耳にBluetoothデバイスを外科手術で埋め込んだ猛者が現る 中国政府が「18歳未満が22時以降にライブ配信を見ること」を禁止へ

中国全土で大学入試期間中にAIサービスが一斉停止 - GIGAZINE

[newstech24]

China shuts down AI instruments throughout nationwide school exams

Chinese language AI corporations have briefly paused a few of their chatbot options to stop college students from utilizing them to cheat throughout nationwide school exams, Bloomberg experiences. In style AI apps, together with Alibaba’s Qwen and ByteDance’s Doubao, have stopped image recognition options from responding to questions on take a look at papers, whereas Tencent’s Yuanbao, Moonshot’s…

[isfeed]

China shuts down AI tools during nationwide college exams

Chinese AI companies have temporarily paused some of their chatbot features to prevent students from using them to cheat during nationwide college exams, Bloomberg reports. Popular AI apps, including Alibaba’s Qwen and ByteDance’s Doubao, have stopped picture recognition features from responding to questions about test papers, while Tencent’s Yuanbao, Moonshot’s Kimi have suspended…

[kamalkafir-blog]

China's AI assistants are getting more handy -- and more invasive

[ASIA] Vivian Toh is chief editor of London-based TechTechChina, a Chinese tech news startup. When ByteDance quietly launched a new screen-sharing function for its AI assistant Doubao in March, the announcement received little fanfare. Yet the technology behind it marks a notable — and controversial — advance in China’s increasingly assertive AI push. if(“undefined”!=typeof…

[aiturtlesai]

SuperGPQA: Bytedance open source benchmark for LLM

  The new SuperGPQA benchmark, developed by experts and the open-source community, offers an assessment of the capabilities of LLMs in 285 disciplines through multiple-choice questions, collaborative filtering and specialized annotations. Key points: Multidisciplinary benchmark with 26,529 multiple-choice questions In-depth assessment in 285 graduate-level disciplines Human-LLM collaborative filtering mechanism supported by expert feedback Methodological directions for future improvement of LLMs ByteDance’s Doubao Large Model Team, in synergy with the open-source M-A-P community,... read more: https://www.turtlesai.com/en/pages-2425/supergpqa-bytedance-open-source-benchmark-for-llm

[beauila-blog]

DeepSeek: The Chinese AI Disruptor Shaking Global Markets

A Chinese artificial intelligence (AI) startup, DeepSeek, has taken the tech world by storm, topping app download charts and triggering a sharp decline in US tech stocks.

In January, the company unveiled its latest model, DeepSeek R1, which it claims matches the capabilities of OpenAI’s ChatGPT while being significantly cheaper to develop. The rapid rise of DeepSeek has unnerved investors, wiping billions off the market value of chip giant Nvidia and challenging the assumption that American firms will dominate the AI industry.

Former US President Donald Trump described DeepSeek’s emergence as a "wake-up call" for American tech companies.

What is Artificial Intelligence?

Artificial intelligence enables machines to learn, analyze data, and solve problems—often mimicking human reasoning. AI has become widely recognized through chatbots like ChatGPT and DeepSeek, which use vast datasets to generate human-like responses.

These AI tools assist users with various tasks, from drafting emails and summarizing text to aiding with coding and academic studies. However, they are also prone to generating misinformation and reflecting biases present in their training data.

Introducing DeepSeek

DeepSeek is a free AI chatbot that functions similarly to ChatGPT. Reports suggest that its R1 model rivals OpenAI’s o1 model in performance, excelling in tasks such as mathematics and coding. Unlike many AI models requiring extensive computing power, DeepSeek R1 was trained at a fraction of the cost—approximately $6 million, compared to the "over $100 million" estimated for OpenAI’s GPT-4.

The company reportedly circumvented US chip restrictions by stockpiling Nvidia A100 chips before the 2022 export ban. Some experts suggest that DeepSeek optimized its resources by integrating these with less sophisticated hardware, making its process more efficient and cost-effective.

These advantages helped DeepSeek's AI assistant become the most downloaded free app on Apple’s App Store in the US. However, its launch was marred by large-scale cyberattacks, leading the company to temporarily limit registrations and contend with website outages.

Like other Chinese AI models—such as Baidu’s Ernie and ByteDance’s Doubao—DeepSeek is programmed to avoid politically sensitive topics. When asked about the Tiananmen Square massacre, for instance, it did not provide details, reflecting China’s strict government censorship policies.

Who is Behind DeepSeek?

DeepSeek was founded in December 2023 by Liang Wenfeng, a graduate of Zhejiang University with degrees in electronic information engineering and computer science. Despite his low public profile, Liang’s influence in the tech and finance sectors is growing rapidly.

He is also the CEO of High-Flyer, a hedge fund specializing in AI-driven quantitative trading. In 2019, High-Flyer became the first quant hedge fund in China to raise over 100 billion yuan ($13 million). Liang has long advocated for China’s AI independence, stating that the country "cannot remain a follower forever."

DeepSeek’s rapid success has drawn global scrutiny. Australia has banned the app on government devices, citing national security concerns. Data protection regulators in several countries have demanded clarity on how the company handles personal information, which is stored on servers in China. Italy went further, blocking the app on January 30 and ordering DeepSeek to cease processing Italian citizens' data.

Impact on US Tech Giants

DeepSeek’s achievements have challenged the belief that AI advancement depends solely on massive budgets and high-end chips. This has created uncertainty in the semiconductor industry, particularly for Nvidia, which saw its stock price plunge 17% before a partial recovery.

The shockwaves sent through financial markets were significant. On January 27, the Nasdaq fell more than 3%, with tech stocks experiencing broad sell-offs. Nvidia, once the world’s most valuable company by market capitalization, dropped to third place behind Apple and Microsoft, with its market value shrinking from $3.5 trillion to $2.9 trillion.

Despite its impact, DeepSeek remains privately owned, meaning investors cannot currently buy shares in the company.

China’s Reaction to DeepSeek’s Rise

DeepSeek’s success is a major boost for the Chinese government, which has been striving for technological self-sufficiency. While Chinese Communist Party officials have not publicly commented, state media has highlighted the app’s impact, describing its rise as a source of anxiety for Silicon Valley and Wall Street.

"DeepSeek’s advances are being celebrated as proof of China’s growing technological prowess and self-reliance," says Marina Zhang, an associate professor at the University of Technology Sydney.

However, Zhang warns that this development could encourage "tech isolationism," as China pushes to further decouple from Western technology.

As the AI race intensifies, DeepSeek’s emergence signals a shift in global power dynamics, proving that innovation is no longer confined to Silicon Valley.

[quotejungle]

Deepseek の AI は、エネルギーとチップの使用を最適化するだけでなく、それ以上のことをする必要に迫られた。たとえば、彼らのモデルはニューロンに似た 6,760 億個のパラメーターしか使用しないが、GPT 4o は 1.8 兆個のパラメーターを誇っている。それでも、出力品質の点では GPT 4o を上回っている。 しかし、Deepseek は、そのような制約下にある唯一の中国の AI モデルではありません。Douyin (およびその国際版 TikTok) の背後にある会社 Bytedance も、いくつかの AI モデルに取り組んでいます。ベンチマーク テストによると、リリースされたばかりの Doubao 1.5 モデルは実行コストが非常に安いため、API 呼び出しに課金しても意味がなく、ユーザーに完全に無料になっています。一方、GPT 4o は 100 万出力トークンあたり 10 ドルを請求します。一部のユーザーはすでに、Deepseek AI の API 呼び出しのコストは 50 セントだけだったのに対し、GPT 4o では 100 ドル以上かかった可能性があると報告しています。Doubao では、このコストは 0 になります。

中国はAI産業が過大評価されていることを一挙に証明

[tumnikkeimatome]

世界第2位のAIチャットボット『ByteDance Doubao（豆包）』ChatGPTに次ぐ中国最強のAI、月間アクティブユーザー数6,000万人、累計ユーザー数1億6,000万人を達成

中国AI市場の覇者…

[datayuan]

Bytedance Seedream 3.0 has caught up with the text-to-image SOTA model GPT-4o, meaning that both Meng and Doubao have been fully launched

[moko1590m]

DeepSeekの優れている点の1つは、オープンソースで公開されているということです。ライバルでクローズドモデルのOpenAIの「o1」とは違い、提供元にお金を払わなくとも好きなサーバーやローカル上で実行することができます。 DeepSeekをオープンソースで公開したことについて、CEOのLiang Wenfeng氏は「オープンソースは人材を惹きつける鍵です。時代を変えるテクノロジーにおいてクローズドソースによって作られた塀は一時的なものであり、OpenAIのクローズドソースのアプローチでさえ、他社の追い上げを防ぐことはできませんでした。私たちは、DeepSeekを通じてノウハウを蓄積し、イノベーションを起こせる組織と文化を形成します。オープンソースは商業的なものというよりも文化的な行動であり、それに貢献することで私たちは尊敬を集められます。今後クローズドソースに変更することはありませんし、まず強力な技術的エコシステムを持つことがより重要だと考えています」と述べています。 DeepSeekのもう1つの優れた点は、AI開発におけるブレークスルーを達成した点です。これまでの一般的な大規模言語モデルは、人間の価値観に沿うように人間がモデルのトレーニングを繰り返す「人間のフィードバックによる強化学習(RLHF)」という手法でトレーニングされ、出力内容に有害なものが生じないよう調整されてきました。 ところが、DeepSeekの「R1-Zero」の方は人間によるフィードバックを削除し、ほとんどの過程で強化学習(RL)のみに基づいたアプローチでトレーニングを行っています。DeepSeekの開発者らによると、「正解」と、「思考プロセスを利用した適切な形式」の2つに対して報酬関数を設定し、段階的に評価しようとしたり可能性のあるすべての答えを検索したりするのではなく、一度に複数の異なる答えを出してから報酬関数に基づく評価を行うようモデルに指示したとのこと。 トンプソン氏は「つまり、推論の仕方を人間が教えずとも、十分な計算量とデータを与えればAIが勝手に推論してくれるという状態を実現したのです」と解説しています。このアプローチによりモデルは思考時間により多くの時間を割り当てることが可能になり、予期せぬ能力が開花しました。これは、従来のデータセットを使い果たすことで学習のスケーリングが限界に近づくという既存のモデルの問題を解決する可能性があります。 DeepSeekはどのようにしてOpenAIの3％のコストでo1を超えたのか？ - GIGAZINE 3つ目の優れた点が、少ないコストで開発されたという点です。OpenAIなどがトレーニングに数十億ドル(数千億円)を費やしていると言われる中、このモデルは600万ドル(約9億3000万円)未満のコストしかかからなかったとされており、またアメリカの半導体輸出規制の中で性能を落として輸出されたNVIDIA製半導体で開発を実現したという点が注目すべきポイントです。 低性能の機器でも高性能の機器に勝るモデルを開発できたという事実により、半導体企業のNVIDIAの価値が揺らいで株価が一時下落しましたが、いずれにせよDeepSeekがNVIDIAの製品に頼ったということに変わりはなく、トンプソン氏は「まだNVIDIAは有利」との見方を示しています。 一方で投資家は頭を悩ませることになります。大企業が多額の費用を投じて進歩を夢見た分野に突然安価でフリーな代替品が現れれば、産業が混乱します。NVIDIA株の下落や、「AIの開発にはエネルギーが必要」との観点から買われていたエネルギー企業の売り転換は、混乱する投資家らの内情がよく反映されていると言えます。海外メディアのAxiosはこの点を「既存の企業に全賭けしたベンチャーキャピタルにとって絶滅レベルの出来事だ」と表現しました。 ただし、DeepSeekが機器の効率的な使用方法を発見したからといって少数の機器で十分だというわけではなく、少なくともNVIDIAの需要はまだ続くことになります。一方でNVIDIA以外の製品で開発することの利点が発見されるなどすれば、そのの地位は揺らぎかねません。 トンプソン氏は「DeepSeekは皆に大きな贈り物を提供したと思います。最大の勝者は消費者と企業であり、長期的にはAIを利用するすべての人が最大の勝者となるでしょう。中国もまた大きな勝者であり、DeepSeekの成功でイノベーションがさらに解き放たれるはずです」と述べました。 この記事のタイトルとURLをコピーする ・関連記事 TikTokのByteDanceがAI「Doubao-1.5-pro」をリリース、OpenAIのモデルに匹敵する中国製AIが続々登場 - GIGAZINE 中国のAI開発企業「DeepSeek」が急速に台頭してテクノロジー業界で話題に、App Storeの無料アプリランキングでも1位を���得 - GIGAZINE DeepSeekのAIモデル「DeepSeek-R1」は天安門広場や台湾、ウイグル族の扱いなどのデリケートな話題に対して中国政府の制限を順守している - GIGAZINE 中国のAI「DeepSeek」ショックでハイテク株がパニック売りに、NVIDIAの時価総額が91兆円消し飛んで暴落記録を2倍以上更新 - GIGAZINE DeepSeekの登場で株価が下落したNVIDIAが「DeepSeekの推論にも相当のNVIDIAチップが必要」と主張し自社の価値をアピール - GIGAZINE DeepSeekが「大規模な悪意のある攻撃」を受けて新規ユーザー登録を制限中 - GIGAZINE サイバーエージェントが「DeepSeek-R1」の派生モデルをベースに日本語で追加学習したモデルを公開 - GIGAZINE ・関連コンテンツ 中国のAI企業・DeepSeekがGPT-4oに匹敵するAIモデル「DeepSeek-V3」をリリース、パラメーター数は脅威の6710億個 話題の中国AI企業「DeepSeek」の創設者兼CEOであるLiang Wenfengへのインタビューからイノベーションの秘訣を探る 2024年にAI分野で何が起きていたのか振り返り、2023年に市場を席捲したGPT-4を上回るAIモデルが多数登場しコンテキストウィンドウの拡張やマルチモーダルモデルが一般化して価格も急落 中国のAI開発企業「DeepSeek」が急速に台頭してテクノロジー業界で話題に、App Storeの無料アプリランキングでも1位を獲得 DeepSeekの登場で株価が下落したNVIDIAが「DeepSeekの推論にも相当のNVIDIAチップが必要」と主張し自社の価値をアピール OpenAI o1相当の推論モデル「DeepSeek R1」を中国AI企業が商用利用や改変が可能なMITライセンスでリリース 中国のAI企業DeepSeekがOpenAI o1に匹敵する推論AIモデル「DeepSeek-R1-Lite-Preview」公開、オープンソース化する計画も 中国のAI「DeepSeek」ショックでハイテク株がパニック売りに、NVIDIAの時価総額が91兆円消し飛んで暴落記録を2倍以上更新

DeepSeekはなぜこんな大騒ぎになっていて一体何がそんなにスゴいのか - GIGAZINE