给ChatGPT和DeepSeek布置了一道我学生的作业题

A
AlexW
楼主 (北美华人网)
我给ChatGPT和DeepSeek布置了这道我学生的作业题:
这是我给它们的打分:
ChatGPT:
DeepSeek:
魏大爷的邻居
好的,用深度求索来解数学题。
d
dmw
不奇怪,美国人数学不好,做出来的AI数学也不会太好,虽然可能有一半的工程师是华人,但抵不住全部华人工程师做的AI解数学题
m
manyworlds
高下立见!
m
minqidev
😂, 现在网上一大堆什么用的ChatGPT的现成的数据的论据。
小东西
你是ChatGPT免费用户,拿着4o/4o mini去跟Deepseek的推理模型比当然比不过啦。
买20块钱的Plus会员就可以用比Deepseek略强的o1模型,买200块钱的Pro会员就可以用到更强的o1 Pro模型。
这里就是盈利和名声之间的博弈。OpenAI想着挣钱,结果强模型不花钱用不到,名声被截胡了。
小东西
蒸馏OpenAI模型的数据当然也是真的,你问Deepseek是谁开发的,部分时候他会回答OpenAI。蒸馏数据跟你用的免费版ChatGPT更不是一个接口,用的是微软Azure上部署的OpenAI model。
y
ynewynew
可是我看不懂啊,没学过
m
minqidev
回复 7楼 的帖子
我就没见过回答openAI的,Deepseek就是免费可以用,不需要钱也能用R1。有什么奇怪的吗? 我就用O1试试。
这个太牛了,也就是说,Deepseek免费版对数学的理解,远高于chat gpt 普通版。 而且比chat gpt o1还要厉害。 第一部分,chat gpt里其实还有个缺陷,没有说明任意多个有线点/线,交集仍为有限。
如果说Chat GPT O1感觉像高考状元,但是DeepSeek就是奥赛冠军,另外一个维度。在数学领域,Chat GPT O1还是会有稍微逻辑缺陷。
小东西
系统prompt会纠正他说自己是OpenAI开发的错误,但是系统prompt不会永远有效。
m
minqidev
回复 10楼 的帖子
其实从数学思维来看,Deepseek 已经超越Chat GPT O1了。他是免费的,还没次数限制,还是开源,
AI竞赛已经结束了,就看光刻机和芯片了 中国只要按照现在的节奏继续培养数学竞赛人才就行
s
sun_flower100
小东西 发表于 2025-01-30 04:27
系统prompt会纠正他说自己是OpenAI开发的错误,但是系统prompt不会永远有效。

Do you have a screenshot of this?
m
minqidev
回复 12楼 sun_flower100 的帖子
截图没用,网上有很多对话框生成器,或者本地版llama,但是用的其他模型。 我现在突然觉得用数学题最对,因为普通不会编造数学题
小东西
基准测试Deepseek-r1还是稍弱于OpenAI o1,更何况o1-pro。明天o3-mini发布,强于o1,说是可以免费使用。
l
lnghng
楼主你这亇比较不对,你这个是ChatGPT,估计是3.5,最新的几个版本是GPT-4o,o1和o3,能力按此顺序依次上升. deepseek的 r1和o1相当,但是需要打开thinking键才是r1,这时候会进行逻辑推理。r1和o1都是有推理的
m
minqidev
回复 15楼 lnghng 的帖子
我用了o1生成结果
k
koala_99
Chatgpt o1 pro:
First, let us restate the problem more explicitly in usual notation. Let R2\Bbb R^2R2 be the plane, and define T = {∅, R2} ∪ {R2∖F ∣ F is a finite union of lines and/or points in R2}.\mathcal{T} \;=\; \bigl\{\varnothing,\, \Bbb R^2\bigr\} \;\cup\; \Bigl\{\Bbb R^2 \setminus F \;\bigm|\; F \text{ is a finite union of lines and/or points in }\Bbb R^2\Bigr\}.T={∅,R2}∪{R2∖F​F is a finite union of lines and/or points in R2}. We want to show: T\mathcal{T}T is indeed a topology on R2\Bbb R^2R2. (R2,T)\bigl(\Bbb R^2, \mathcal{T}\bigr)(R2,T) is a T1T_1T1​-space. (R2,T)\bigl(\Bbb R^2, \mathcal{T}\bigr)(R2,T) is not Hausdorff. Below is a concise proof of each claim. 1. T\mathcal{T}T is a topology To verify that T\mathcal{T}T is a topology, we need to check: ∅∈T\varnothing \in \mathcal{T}∅∈T and R2∈T\Bbb R^2 \in \mathcal{T}R2∈T. These are included by definition. T\mathcal{T}T is closed under arbitrary unions. The usual (and simpler) way to see this is to view {R2∖F ∣ F is finite union of lines/points} \Bigl\{\Bbb R^2 \setminus F \;|\;F \text{ is finite union of lines/points}\Bigr\}{R2∖F∣F is finite union of lines/points} as a subbase for T\mathcal{T}T. The finite intersections of these subbase elements give complements of finite unions of lines/points (since intersection of two complements is the complement of the union). Then T\mathcal{T}T is the collection of all unions of these finite intersections, making it the coarsest topology containing all such complements. By construction, it is closed under arbitrary unions. T\mathcal{T}T is closed under finite intersections. If U1=R2∖F1U_1 = \Bbb R^2 \setminus F_1U1​=R2∖F1​ and U2=R2∖F2U_2 = \Bbb R^2 \setminus F_2U2​=R2∖F2​, where each FiF_iFi​ is a finite union of lines and/or points, then U1∩U2 = (R2∖F1) ∩ (R2∖F2) = R2 ∖ (F1∪F2). U_1 \cap U_2 \;=\; (\Bbb R^2 \setminus F_1) \;\cap\; (\Bbb R^2 \setminus F_2) \;=\; \Bbb R^2 \;\setminus\; \bigl(F_1 \cup F_2\bigr).U1​∩U2​=(R2∖F1​)∩(R2∖F2​)=R2∖(F1​∪F2​). Since F1∪F2F_1 \cup F_2F1​∪F2​ is again a finite union of lines/points, U1∩U2∈TU_1 \cap U_2 \in \mathcal{T}U1​∩U2​∈T. Similarly for any finite intersection. Hence T\mathcal{T}T satisfies the axioms of a topology. 2. T\mathcal{T}T is T1T_1T1​ A topological space is T1T_1T1​ if and only if, for every pair of distinct points x≠yx \neq yx=y, there is an open set containing xxx that does not contain yyy. Given distinct points x,y∈R2x,y \in \Bbb R^2x,y∈R2, consider the set U = R2 ∖ {y}. U \;=\; \Bbb R^2 \,\setminus\, \{y\}.U=R2∖{y}. This is the complement of a finite set of points (just the single point {y}\{y\}{y}), hence U∈TU \in \mathcal{T}U∈T. Clearly x∈Ux \in Ux∈U but y∉Uy \not\in Uy∈U. Therefore, T\mathcal{T}T satisfies the T1T_1T1​ separation axiom.
3. T\mathcal{T}T is not Hausdorff A topology is Hausdorff (or T2T_2T2​) if for every pair of distinct points x,yx,yx,y, there exist disjoint open sets U,VU,VU,V with x∈Ux \in Ux∈U and y∈Vy \in Vy∈V. We will see this fails in (R2,T)\bigl(\Bbb R^2, \mathcal{T}\bigr)(R2,T). Suppose x≠yx \neq yx=y. Let UUU be any open set containing xxx. By definition, UUU is either all of R2\Bbb R^2R2 or of the form R2∖F\Bbb R^2 \setminus FR2∖F, where FFF is a finite union of lines/points. If we also pick VVV open containing yyy, then V=R2∖GV = \Bbb R^2 \setminus GV=R2∖G for some finite union GGG of lines/points. The intersection U∩VU \cap VU∩V equals (R2∖F) ∩ (R2∖G) = R2 ∖ (F∪G). \bigl(\Bbb R^2 \setminus F\bigr) \;\cap\; \bigl(\Bbb R^2 \setminus G\bigr) \;=\; \Bbb R^2 \;\setminus\; (F \cup G).(R2∖F)∩(R2∖G)=R2∖(F∪G). But F∪GF \cup GF∪G is still a finite union of lines and/or points. A finite union of lines and points in R2\Bbb R^2R2 can never cover the entire plane. Hence R2∖(F∪G) ≠ ∅. \Bbb R^2 \setminus (F \cup G) \;\neq\; \varnothing.R2∖(F∪G)=∅. In other words, any two open sets in T\mathcal{T}T intersect in a nonempty set. There simply is no way to choose UUU and VVV so that U∩V=∅U \cap V = \varnothingU∩V=∅. Thus (R2,T)\bigl(\Bbb R^2, \mathcal{T}\bigr)(R2,T) is not Hausdorff. Conclusion We have shown T\mathcal{T}T is indeed a topology (it is generated by taking complements of a finite collection of lines/points). It is T1T_1T1​ because single points can be excluded by an open set. It is not Hausdorff, since no two open sets in this topology can be disjoint (finite unions of lines/points never fill the plane). Hence all parts of the statement are proved.
s
shanggj
koala_99 发表于 2025-01-30 16:30
Chatgpt o1 pro:
First, let us restate the problem more explicitly in usual notation. Let R2\Bbb R^2R2 be the plane, and define T = {∅, R2} ∪ {R2∖F ∣ F is a finite union of lines and/or points in R2}.\mathcal{T} \;=\; \bigl\{\varnothing,\, \Bbb R^2\bigr\} \;\cup\; \Bigl\{\Bbb R^2 \setminus F \;\bigm|\; F \text{ is a finite union of lines and/or points in }\Bbb R^2\Bigr\}.T={∅,R2}∪{R2∖F​F is a finite union of lines and/or points in R2}. We want to show: T\mathcal{T}T is indeed a topology on R2\Bbb R^2R2. (R2,T)\bigl(\Bbb R^2, \mathcal{T}\bigr)(R2,T) is a T1T_1T1​-space. (R2,T)\bigl(\Bbb R^2, \mathcal{T}\bigr)(R2,T) is not Hausdorff. Below is a concise proof of each claim. 1. T\mathcal{T}T is a topology To verify that T\mathcal{T}T is a topology, we need to check: ∅∈T\varnothing \in \mathcal{T}∅∈T and R2∈T\Bbb R^2 \in \mathcal{T}R2∈T. These are included by definition. T\mathcal{T}T is closed under arbitrary unions. The usual (and simpler) way to see this is to view {R2∖F ∣ F is finite union of lines/points} \Bigl\{\Bbb R^2 \setminus F \;|\;F \text{ is finite union of lines/points}\Bigr\}{R2∖F∣F is finite union of lines/points} as a subbase for T\mathcal{T}T. The finite intersections of these subbase elements give complements of finite unions of lines/points (since intersection of two complements is the complement of the union). Then T\mathcal{T}T is the collection of all unions of these finite intersections, making it the coarsest topology containing all such complements. By construction, it is closed under arbitrary unions. T\mathcal{T}T is closed under finite intersections. If U1=R2∖F1U_1 = \Bbb R^2 \setminus F_1U1​=R2∖F1​ and U2=R2∖F2U_2 = \Bbb R^2 \setminus F_2U2​=R2∖F2​, where each FiF_iFi​ is a finite union of lines and/or points, then U1∩U2 = (R2∖F1) ∩ (R2∖F2) = R2 ∖ (F1∪F2). U_1 \cap U_2 \;=\; (\Bbb R^2 \setminus F_1) \;\cap\; (\Bbb R^2 \setminus F_2) \;=\; \Bbb R^2 \;\setminus\; \bigl(F_1 \cup F_2\bigr).U1​∩U2​=(R2∖F1​)∩(R2∖F2​)=R2∖(F1​∪F2​). Since F1∪F2F_1 \cup F_2F1​∪F2​ is again a finite union of lines/points, U1∩U2∈TU_1 \cap U_2 \in \mathcal{T}U1​∩U2​∈T. Similarly for any finite intersection. Hence T\mathcal{T}T satisfies the axioms of a topology. 2. T\mathcal{T}T is T1T_1T1​ A topological space is T1T_1T1​ if and only if, for every pair of distinct points x≠yx \neq yx=y, there is an open set containing xxx that does not contain yyy. Given distinct points x,y∈R2x,y \in \Bbb R^2x,y∈R2, consider the set U = R2 ∖ {y}. U \;=\; \Bbb R^2 \,\setminus\, \{y\}.U=R2∖{y}. This is the complement of a finite set of points (just the single point {y}\{y\}{y}), hence U∈TU \in \mathcal{T}U∈T. Clearly x∈Ux \in Ux∈U but y∉Uy \not\in Uy∈U. Therefore, T\mathcal{T}T satisfies the T1T_1T1​ separation axiom.
3. T\mathcal{T}T is not Hausdorff A topology is Hausdorff (or T2T_2T2​) if for every pair of distinct points x,yx,yx,y, there exist disjoint open sets U,VU,VU,V with x∈Ux \in Ux∈U and y∈Vy \in Vy∈V. We will see this fails in (R2,T)\bigl(\Bbb R^2, \mathcal{T}\bigr)(R2,T). Suppose x≠yx \neq yx=y. Let UUU be any open set containing xxx. By definition, UUU is either all of R2\Bbb R^2R2 or of the form R2∖F\Bbb R^2 \setminus FR2∖F, where FFF is a finite union of lines/points. If we also pick VVV open containing yyy, then V=R2∖GV = \Bbb R^2 \setminus GV=R2∖G for some finite union GGG of lines/points. The intersection U∩VU \cap VU∩V equals (R2∖F) ∩ (R2∖G) = R2 ∖ (F∪G). \bigl(\Bbb R^2 \setminus F\bigr) \;\cap\; \bigl(\Bbb R^2 \setminus G\bigr) \;=\; \Bbb R^2 \;\setminus\; (F \cup G).(R2∖F)∩(R2∖G)=R2∖(F∪G). But F∪GF \cup GF∪G is still a finite union of lines and/or points. A finite union of lines and points in R2\Bbb R^2R2 can never cover the entire plane. Hence R2∖(F∪G) ≠ ∅. \Bbb R^2 \setminus (F \cup G) \;\neq\; \varnothing.R2∖(F∪G)=∅. In other words, any two open sets in T\mathcal{T}T intersect in a nonempty set. There simply is no way to choose UUU and VVV so that U∩V=∅U \cap V = \varnothingU∩V=∅. Thus (R2,T)\bigl(\Bbb R^2, \mathcal{T}\bigr)(R2,T) is not Hausdorff. Conclusion We have shown T\mathcal{T}T is indeed a topology (it is generated by taking complements of a finite collection of lines/points). It is T1T_1T1​ because single points can be excluded by an open set. It is not Hausdorff, since no two open sets in this topology can be disjoint (finite unions of lines/points never fill the plane). Hence all parts of the statement are proved.

LZ 能出来打下分吗?
现在看来 大学本科低年级数学专业 的数学作业, AI 都可以了。