人工智能历史1956DartmouthConference:TheFoundingFathers of Al标志性事件：1956年的Dartmouth会议，十人左右参加，历时两个月。当时创立的计算机系通常分三个领域：理论，系统，AIAI领域显示度较高的项目：下棋。从1957年开始，每十年都预言十年内机器能下赢国际象棋世界冠军。一直等了40年，到1997才真正实现。量变到质变：1995年卡斯帕罗夫批评计算机下棋没有悟性/直觉(Insights）：1997年失败的时候还在猜是不是计算机作弊（有人类象棋大师群在背后集体决策）从这里可以推论悟性/直觉指的是什么机器会思考吗？E乔姆斯基回答“潜艇会游泳吗？
人工智能历史 标志性事件：1956年的 Dartmouth 会议，十人左右参加，历时两个月。 当时创立的计算机系通常分三个领域：理论，系统，AI。 AI领域显示度较高的项目：下棋。从1957年开始，每十年都预言十年内 机器能下赢国际象棋世界冠军。一直等了40年，到1997才真正实现。 • 量变到质变：1995年卡斯帕罗夫批评计算机下棋没有悟性/直觉 (Insights)；1997年失败的时候还在猜是不是计算机作弊（有人类象棋大 师群在背后集体决策） • 从这里可以推论悟性/直觉指的是什么 机器会思考吗？ 乔姆斯基回答 “潜艇会游泳吗？
思考题，在chatGPT可以进行类人创作和问答之后，还有什么是智能体/大模型理论上做不到的？23
▪ 在chatGPT 可以进行类人创作和问答之后，还有什么是智能体/大模型 理论上做不到的？ ▪ 1 ▪ 2 ▪ 3 思考题
Automated Theorem Provingsm- Russell & Whitehead, 1910s, Principia Mathematica, derive allmathematical truth using axioms and inference rules of formal*5443.i.a.Bel.D:anBmA.=.aupe2logic.Dem.F.*5426.ofi.awta.p=ty.D:aupe2.#.ay.[51-231]=.t'anty=A“Logic Theorist"could provea[*13-12](1)=.anp-A.(1).*11:11-35.)subset in Principia Mathematica.(2)Fi.(az,y).a-ta.Bmty.D:aue2.=.anp=AF.(2).*11-54.*52-1.DF.PropFrom this proposition it will follow, when arithmetical addition has been1959, Hao Wang provedalldefined, that 1+1=2.theorems in PrincipiaMathematicaon an IBM 704.1976,fourcolortheorem.- 2016, Boolean Pythagorean Triple Problem, 200TB. There are10^2355 possible solutions, compare with 10^80 atoms in theuniversity and 10^18 seconds since the big bang
▪ Russell & Whitehead, 1910s, Principia Mathematica, derive all mathematical truth using axioms and inference rules of formal logic. ▪ “Logic Theorist” could prove a subset in PrincipiaMathematica. ▪ 1959, Hao Wang provedall theorems in Principia Mathematica on an IBM 704. ▪ 1976, four color theorem. ▪ 2016, Boolean Pythagorean Triple Problem, 200TB. There are 10^2355 possible solutions, compare with 10^80 atoms in the university and 10^18 seconds since the big bang. Automated Theorem Proving
ExpertSystem1960s,thefirstexpert system,DENTRAL1970s,MYCIN,identify bacteriacausing severeinfectionsandrecommendantibiotics1980s, XCON, assist in the ordering ofDEC's VAXcomputersystems by automatically selecting the computer systemcomponentsbasedonthecustomer'srequirementsNow still usedin credit investigation,risk management, businessrules engine,etc
▪ 1960s, the first expert system,DENTRAL. ▪ 1970s, MYCIN, identify bacteria causing severe infectionsand recommend antibiotics. ▪ 1980s, XCON, assist in the ordering of DEC's VAXcomputer systems by automatically selecting the computer system components based on the customer's requirements. ▪ Now still used in credit investigation, risk management, business rules engine, etc. Expert System
Knowledge RepresentationLogics:(Vx)Man(x)Mortal(x) & Man(Socrates) → Mortal(Socrates)WordNetSearch-3.1WordNethome fageClossary-HeipWord to.search for:.HamanSearch.WordietDrsplayOptions(Seiect opban techange)ChangShowWord(leccal)reltotKrSShow:Synsetisepiayopitioms.torsensn.(glossj"anexamgiesertoncoWordNet:NounS(n)momo.man,humanboing.human (any living orextinctmemberot thefamtyaecharactenzedtvsupernorntellonce.arfculalesoeethanderectcamageAdjective(adi)human (charactoristicofhumanity)humannature(adhuman (relatingtoaperson)zheexperimentwasconductedon.6monkeysand2humans(ad)human (navinghumanform orattributes as opposedto thoseofanimais ordivinebengs)humanbeings"hehuwnanbodyhmankndness"hcmantfraityConceptualgraph,SemanticwebKnowledgegraph
▪ Logics: ∀𝑥 𝑀𝑎𝑛 𝑥 ⊃ 𝑀𝑜𝑟𝑡𝑎𝑙 𝑥 & 𝑀𝑎𝑛(𝑆𝑜𝑐𝑟𝑎𝑡𝑒𝑠) → 𝑀𝑜𝑟𝑡𝑎𝑙(𝑆𝑜𝑐𝑟𝑎𝑡𝑒𝑠) ▪ WordNet: ▪ Conceptual graph, Semanticweb ▪ Knowledge graph Knowledge Representation