→ cs09312: 所以結論是chatgpt抄DS嗎 39.9.67.159 02/06 18:55
※ 編輯: feelthat (114.32.93.159 臺灣), 02/06/2025 19:06:11
→ nekoares: 「AI沒有錯!(笑)」 111.248.102.98 02/06
16:32
本版不缺 稻草人論證 99.99% 都如此
不見得AI有錯!!! -> 「AI沒有錯!(笑)」
某問:支持妓女有權做生意嗎? 我回:支持! -> 你一定去嫖過妓
推 SShirakawa: 建構式數學加上文字邏輯...好像台灣現 114.136.129.162 02/06 16:34
→ SShirakawa: 在在教小朋友的那套啊 114.136.129.162 02/06 16:34
////
推 stratuss: DeepSeek完全沒問題 220.137.205.44 02/06 17:31
→ feelthat: DeepSeek一開始就以中文編碼中文問剛好 114.32.93.159 02/06 18:23
這裡有簡易版的gpt, bert chatbot model 稍微懂一點也就了解AI問題出在那了
https://tinkerd.net/blog/machine-learning/bert-embeddings/
//
會有這些問題的人 因為沒去理解chat ai model原理才會發生的
那說原因了 還有人不以為然 chatbot AI基本運作知識不足外 惡意嘲諷
現在我以另一種語文去問deepseek 答案與chatgpt 一樣
https://www.mediafire.com/view/kb5ifrk0xsvuy12/exdai00.png/file
https://www.mediafire.com/view/wytqrtwxei1zf7f/exdai01.png/file
所以可以知道deepseek的運作方式 與 chatgpt 應該有相似之處
※ 引述《feelthat (雲端傳送)》之銘言：
: AI先用語言編碼再進神經網路最後歸納演算
: 請用英文問且 關鍵字 要正確
: calculate math 9.11-9.9 is?
: ChatGPT 說：
: The result of
: 9.11
: 9.9
: 9.11.9 is:
: 9.11
: 9.9
: =
: 0.79
: 9.11.9=-.79
: //另一段
: I mean 9.11 - 9.9 = -0.79 but now 9.11 - 9.9 = 0.21
: maybe it's not decimal maybe some situation can you list?
: ChatGPT 說：
: Ah, I get it now! You're asking for scenarios where the result of
: 9.11
: 9.9
: 9.11.9 could change from
: 0.79
: .79 to
: 0.21
: 0.21, which might not involve decimal arithmetic. Here's a list of
: possibilities where such a situation could occur:
: 1. Modulo Arithmetic
: In modular arithmetic, results "wrap around" after reaching a certain value.
: For example:
: If we are working in modulo
: 1
: 1,
: 9.11
: 9.11 becomes
: 0.11
: 0.11 and
: 9.9
: 9.9 becomes
: 0.9
: 0.9.
: Subtraction
: 0.11
: 0.9
: 0.11.9 in modulo
: 1
: 1 gives
: 0.21
: 0.21 (wrapping back around).
: 2. Custom Number Systems or Scales
: The subtraction is performed in a non-decimal number system, such as:
: Base-12 (duodecimal): Numbers in different bases can yield results that don't
: align with decimal subtraction.
: Arbitrary scale adjustment: For example,
: 9.11
: 9.11 and
: 9.9
: 9.9 are values scaled by a factor or offset, changing the subtraction outcome.
: 3. Signed vs. Unsigned Interpretation
: If the numbers are stored as signed vs. unsigned integers in computer systems:
: 9.11
: 9.9
: 9.11.9 could behave differently depending on how the system interprets
: values.
: Overflow or underflow in unsigned subtraction could result in a different
: outcome.
: 4. Error Correction or Adjustment Factor
: The calculation might include a hidden correction factor or adjustment:
: Some systems might add a constant adjustment to ensure positive results.
: For example, a constant
: +
: 1
: +1 correction:
: (
: 9.11
: 9.9
: )
: +
: 1
: =
: 0.21
: (9.11.9)+1=0.21
: 5. Quantum or Probabilistic Systems
: In quantum computing or probabilistic algorithms:
: Subtraction results might not be deterministic, allowing
: 9.11
: 9.9
: 9.11.9 to
: 0.21
: 0.21 in
: 6. Time-Based Context
: In specific contexts like timestamps:
: 9.11
: 9.11 and ( 9.9
: 9.9
: 9.9 might represent
: 7. Measurement Errors or Bias
: In physical systems, results could change due to:
: Mea
: Bi
: //
: 9.11 - 9.9 = -0.79 but now 9.11 - 9.9 = 0.21
: maybe it's not decimal maybe some situation can you list?
: ChatGPT 說：
: Great catch! If
: 9.11
: 9.9
: =
: 0.21
: 9.11.9=0.21, then the numbers might not be in decimal form. Let's explore
: potential scenarios where this could make sense:
: 1. Base Systems (Non-Decimal)
: Base 10 (Decimal):
: 9.11
: 9.9
: =
: 0.79
: 9.11.9=.79, as calculated initially.
: Base 12 (Duodecimal):
: 9.11
: 9.11 and
: 9.9
: 9.9 are interpreted in base 12.
: 9.11
: 12
: =
: 9
: +
: 1
: 12
: +
: 1
: 144
: 9.11
: 12
: =9+
: 12
: 1
: +
: 144
: 1
: 9.9
: 12
: =
: 9
: +
: 9
: 12
: 9.9
: 12
: =9+
: 12
: 9
: Subtraction could yield
: 0.21
: 0.21 when converted to base 10.
: 2. Time Format (Hours and Minutes)
: Hours and Minutes:
: 9.11
: 9.11 could mean 9 hours and 11 minutes.
: 9.9
: 9.9 could mean 9 hours and 90 minutes (or 10 hours and 30 minutes).
: Subtracting
: 9
: 11
: 10
: 30
: 9:110:30 might give you
: 0
: 21
: 0:21 (21 minutes).
: 3. Measurement Systems
: If
: 9.11
: 9.11 and
: 9.9
: 9.9 are part of a measurement system with specific units or rounding
: conventions, the difference might result in
: 0.21
: 0.21.
: 4. Financial Rounding or Precision
: In accounting or financial settings, numbers might be rounded or truncated
: based on significant figures, giving a result of
: 0.21
: 0.21.
: //
: 是我們想得太簡單 不見得AI有錯!!!
: ※ 引述《bengowa (鞭狗蛙 舒服)》之銘言：
: : 乳題啦
: : 我剛剛看到有人PO文說
: : 如果你去問chatgpt一個簡單的減法問題
: : 9.11-9.9=?
: : 他會跟你說0.21
: : 我本來想說 幹怎麼可能
: : 結果測試後真的是這樣
: : https://imgur.com/6cQOyFC.jpg
: : 而且解釋後重複問他 他還是答錯
: : 笑死==
--
※ 發信站: 批踢踢實業坊(ptt.cc), 來自: 114.32.93.159 (臺灣)
※ 文章網址: https://www.ptt.cc/bbs/Gossiping/M.1738839150.A.C42.html