儘管大肆宣傳，ChatGPT-4o 還是表現平平

狗狗幣創始人 Billy Markus（又名 Shibetoshi Nakamoto）透露了 OpenAI 的 ChatGPT-4o 的即時結果，並對其表現表示不滿。儘管是 OpenAI 最新、最複雜的聊天機器人，Markus 證明 ChatGPT-4o 對有關 GME 股票提示的響應不準確，並且與預期輸出相去甚遠。

ChatGPT-4o Reveals Imperfections, Despite Grand Promises

儘管前景廣闊，但 ChatGPT-4o 仍有缺陷

Dogecoin co-founder Billy Markus, widely known as Shibetoshi Nakamoto on social media platforms, has recently disclosed the results of a prompt request made to OpenAI's latest chatbot, ChatGPT-4o. Markus shared that he instructed the AI to "make a meme about GME stock," expecting a witty or humorous response. However, the result left Markus unimpressed and prompted him to question the capabilities of this highly anticipated language model.

狗狗幣聯合創始人 Billy Markus（在社交媒體平台上被廣泛稱為 Shibetoshi Nakamoto）最近披露了向 OpenAI 最新聊天機器人 ChatGPT-4o 發出的即時請求的結果。馬庫斯表示，他指示人工智慧“製作一個關於 GME 股票的表情包”，期待得到機智或幽默的回應。然而，結果讓馬庫斯不以為然，並促使他質疑這種備受期待的語言模式的能力。

Markus's prompt was carefully chosen, considering the recent resurgence of interest in GameStop stock. The gaming retailer's shares had surged by over 80% since the preceding Friday's close, making it a topical and relevant subject for a meme. As an advanced LLM engine, ChatGPT-4o was expected to effortlessly navigate the prompt by drawing upon its vast knowledge and ability to understand context.

考慮到最近人們對 GameStop 股票的興趣重新抬頭，馬庫斯的提示是經過精心挑選的。自上週五收盤以來，這家遊戲零售商的股價已飆升超過 80%，使其成為熱門話題。作為一個先進的 LLM 引擎，ChatGPT-4o 預計將利用其豐富的知識和理解上下文的能力，輕鬆地導航提示。

Disappointingly, the result generated by ChatGPT-4o fell far short of Markus's expectations. The chatbot's response was neither humorous nor insightful, instead producing a rather bland and uninspired meme that simply stated "crayon eater." Markus's tweet, which included a screenshot of the exchange, sparked a lively discussion among followers, highlighting the perceived limitations of the supposedly cutting-edge chatbot.

令人失望的是，ChatGPT-4o 產生的結果遠低於 Markus 的預期。聊天機器人的反應既不幽默，也不富有洞察力，而是產生了一個相當平淡無奇的模因，只是簡單地寫著「蠟筆食者」。馬庫斯的推文包含了這次交流的螢幕截圖，引發了追蹤者的熱烈討論，強調了所謂的尖端聊天機器人的局限性。

Markus's experience with ChatGPT-4o raises concerns about the true capabilities of LLMs, despite their widely publicized advancements. While these models have demonstrated impressive abilities in certain areas, such as generating text, translating languages, and answering questions, they still exhibit significant shortcomings when it comes to understanding and responding to complex prompts or generating nuanced content.

馬庫斯在 ChatGPT-4o 的經驗引起了人們對法學碩士真實能力的擔憂，儘管它們的進步已被廣泛宣傳。儘管這些模型在某些領域（例如生成文字、翻譯語言和回答問題）表現出了令人印象深刻的能力，但在理解和回應複雜提示或產生細緻入微的內容方面仍然表現出明顯的缺點。

The shortcomings of ChatGPT-4o serve as a reminder that LLMs are still very much in their infancy, and their development is an ongoing process. While they have the potential to revolutionize various industries, their limitations must be acknowledged and addressed. As researchers and developers continue to refine and improve these models, it is crucial to maintain realistic expectations about their capabilities and to approach their results with a critical eye.

ChatGPT-4o 的缺點提醒人們，法學碩士仍處於起步階段，其發展是一個持續的過程。雖然它們有潛力徹底改變各個行業，但必須承認並解決它們的局限性。隨著研究人員和開發人員不斷改進和改進這些模型，保持對其能力的現實期望並以批判的眼光對待他們的結果至關重要。