將任何主題轉化為專業且有來源佐證的研究報告,透過多代理進行深入研究。
Deep Research 能將一個主題轉化為達到機構級水準的研究報告——這種結構完整、來源明確的文件,正是專業研究團隊會撰寫的內容,而非幾份搜尋摘要。
提供研究主題後,多代理工作流程便會接手:規劃大綱、透過分層資料來源平行搜尋、擷取並查證資料、並行撰寫各章節,最後在彙整前進行內建品質檢查。編輯標準嚴格:每章開頭都會提出核心判斷;每項數據都附有可追溯的來源引註;每份報告都會在主要論點之外呈現相反觀點;分析則分為三個層次——事實、因果關係與判斷——不添加冗文。標準流程會產出包含數十個資料表、並引用 15-25 個不重複來源的報告;快速與深入模式則可依需求調整深度與速度。
它會自動偵測主題所使用的語言,並以 19 種語言進行原生研究,直接以報告目標語言搜尋來源,而非事後翻譯。適用於產業與市場研究、競爭分析、政策分析、技術深入研究、趨勢預測和投資備忘錄;完成的報告可匯出為 PDF 或 Word 文件,方便分享。
從英文翻譯,僅供參考。
According to the current academic consensus, when is artificial general intelligence (AGI) expected to be achieved?
There is no academic consensus on a date for AGI. As of October 2026, it would be misleading to say that researchers generally agree it will arrive in 2030, 2040, or any other particular year.
The most useful distinction is between expert forecasts and scientific consensus: surveys tell us what researchers expect, not what has been established scientifically.
A large survey of 2,778 researchers who had published in leading AI venues, conducted in October 2023 and subsequently published in the Journal of Artificial Intelligence Research, reported these aggregate forecasts:
These forecasts assumed scientific progress continued without disruption. The 50% date for the first milestone moved from 2060 in the 2022 survey to 2047 in the 2023 survey, illustrating how quickly expectations can shift. Importantly, these are 2023 opinions—not an updated 2026 forecast, despite the paper’s later publication.
Bottom line: If you want a numerical reference point, 2047 was the 50%-probability date in a major researcher survey for a particularly demanding human-level-AI milestone. But the academically defensible answer is “the timing remains highly uncertain,” not “AGI is expected in 2047.” Neither “within a few years” nor “several decades away” is an agreed academic conclusion.
本次執行費用104 點數≈ $0.104網路搜尋 ×2GPT-6.1 Sol · 49,025 詞元
執行 AI 代理會消耗你的點數。隨時都能停止——你只需為它實際產出的內容付費。