Select Page

Original & Concise Bullet Point Briefs

How ChatGPT changes the future of programming | Stephen Wolfram and Lex Fridman

Computational Accessibility Opens Doors for Non-Programmers

  • The democratization of access to computation is becoming more accessible due to the development of linguistic interfaces
  • Wolfman Alpha is an example of a language model that allows users to access deep computation
  • Boilerplate programming is becoming less necessary as higher-level languages are developed
  • With this new technology, art history students and other non-programmers can now use it before having to learn it.

Investigating What University Subjects are Most Valuable: The Use of Computational Language and Expository Writing for Artistic Programs

  • People will learn how to use computational language to control programs
  • People must also understand the architecture and possibilities of what is computationally possible
  • There is an artistic element to controlling the car using natural language, not just knowing where you want to go
  • Expository writing could be a useful skill for writing prompts
  • A study has been undertaken investigating what fields of knowledge have been considered worth learning by universities in the past, such as geography and linguistics.

Exploring the Intersection of AI, Computer Science and Security: Mind Hacks and Jailbreaking LLMs

  • Computer Science is the representation of how we think about and represent the world computationally
  • AI has grown to be able to understand humans in a remarkable way, utilizing expository mechanisms and thought experiments
  • AI Wranglers (psychotherapists) use manipulative or game theoretic interaction with AI to get a deep truth
  • Computer security aspects such as phishing can be used for LLMs
  • Mind Hacks are similar between humans and LLMs though there may be “jailbreaking techniques” for LLMs not yet known.

Computational Language: Bridging the Gap Between Natural and Programmed Language

  • Computational thinking is a formal way of looking at the world, allowing for the building of a tower of capabilities
  • Natural language may evolve to combine aspects of computational language
  • There is incentive for young people to learn spoken computational language
  • It should be possible to convert computational language into a spoken language with dictation being easy
  • Human language has tricks that could be used for spoken computational language.

Colleges of Computing Adapt Curriculum to Keep Pace With Technological Advancements

  • College of Computing will likely focus on teaching computational thinking, understanding data science, and how to formalize or “computationalize” the world
  • This includes teaching concepts such as digital data representation, statistical analysis, aggregated preferences and average calculations
  • In order to do so, students should be taught about programming languages, color spaces, optical concepts like lenses and chromatic aberration, hue saturation brightness space and RGB space
  • The field is evolving quickly due to new technologies like MLMs
  • Therefore colleges must continue to adapt their curricula
  • One project is being undertaken by someone to create a textbook that explains fundamentals of computer science such as bugs and software testing.

Universities Approach Teaching Mathematics and Computational Literacy to Students: Need for a Single Course to Bridge Subjects and Technical Writing

  • Computational literacy is important in all fields of study
  • Universities have different approaches to teaching mathematics
  • All subjects assume some level of writing literacy, though it is not taught separately in each department
  • Technical writing requires a certain level of understanding
  • There is a need for a single course to provide computational literacy to students that can be applied to any subject
  • It may take a year long college class to reach adequate levels of computational literacy.

Original & Concise Bullet Point Briefs

With VidCatter’s AI technology, you can get original briefs in easy-to-read bullet points within seconds. Our platform is also highly customizable, making it perfect for students, executives, and anyone who needs to extract important information from video or audio content quickly.

  • Scroll through to check it out for yourself!
  • Original summaries that highlight the key points of your content
  • Customizable to fit your specific needs
  • AI-powered technology that ensures accuracy and comprehensiveness
  • Scroll through to check it out for yourself!
  • Original summaries that highlight the key points of your content
  • Customizable to fit your specific needs
  • AI-powered technology that ensures accuracy and comprehensiveness

Unlock the Power of Efficiency: Get Briefed, Don’t Skim or Watch!

Experience the power of instant video insights with VidCatter! Don’t waste valuable time watching lengthy videos. Our AI-powered platform generates concise summaries that let you read, not watch. Stay informed, save time, and extract key information effortlessly.

can you kind of give your wise Sageadvice about what humans who have neverinteracted with the eye systemsuh not even like with wolf from Alphaare now interacting with Chad GPTbecause it becomesit's accessible to a certain demographicthey may have not touched AI systemsbefore what do we do with truth likejournalists for exampleyeah how do we think about the output ofthese systems I think this ideathe idea that you're going to getfactual output is not a very good idea Imean it's just this is not it is alinguistic interface it is producinglanguage and language can be truthful ornot truthful and that's a differentslice of what's going on I think thatyou know what we seeand for example uh kind of you know gocheck this with your fact source forexample you can do that to some extentbut then it's going to not checksomething it's going you know that isagain a thing that is sort of a does itcheck in the right place I mean we seethat in you know does it call the youknow the Wolfman plug-in in the rightplace you know often it does sometimesit doesn't you know I I think the thereal thing to understand about what'shappening is which I think is veryexciting is kind of the the greatdemocratization of access to computationyeah and and um you know I think thatwhen you look at sort of the there'sbeen a long period of time whencomputation and the ability to figureout things with computers has beensomething that kind of only the only TheDruids at some level can can achieve youknow I myself have been involved intrying to sort of de-druidifyum access to computation I mean backbefore Mathematica existed you know in1988 if you were a you know a physicistor something like that and you wanted todo a computation you would find aprogrammer you would go and you knowdelegate the the computation to thatprogrammer hopefully they'd come backwith something useful maybe theywouldn't there'd be this long you knowmulti-week you know Loop that you gothrough and then it was actually veryvery interesting to see 1988 you knowlike first people like physicistsmathematicians and so on then other lotsof other people but this very rapidtransition of people realizing theythemselves could actually type withtheir own fingers and you know make somepiece of code that would do acomputation that they cared about andyou know it's been exciting to see lotsof discoveries and so on made by byusing that tool and I think the samething is you know we see the same thingyou know Wolfman Alpha is dealing withuh is not as deep computation as you canachieve with whole woven languageMathematica stack but the thing that'sto me particularly exciting about kindof the large language model linguisticinterface mechanism is it dramaticallybroadens the access to kind of deepcomputation I mean it's kind of like oneof the things I've sort of thought aboutrecently is you know what's going tohappen to all these programmers what'sgoing to happen to all these people whoyou know a lot of what they do is writeslabs of boilerplate code and in a senseyou know I've been saying for 40 yearsthat's not a very good idea you know youcan automate a lot of that stuff with ahigh enough level language that slab ofcode that's designed in the right wayyou know that slab of code turns intothis one function we just implements ityou can just useum so in a sense that the fact thatthere's there's all of this activity ofdoing sort of lower level programming issomething for me it seemed like I don'tthink this is the right thing to do butyou know and and lots of people haveused our technology and not had to dothat but the fact is that that's youknow so when you look at I don't knowcomputer science departments that havethat have turned into places wherepeople are learning the trade ofprogramming so to speak it's it's sortof a question of what's going to happenand I think there are two Dynamics oneis that kind of uh sort of uhboilerplate programming is going tobecome you know it's going to go the waythat Assembly Language went back in theday of something where it's reallymostly specified by at a higher levelyou know you start with natural languageyou turn it into a computationallanguage that's you look at thecomputational language you run tests youunderstand that's what's supposed tohappen you know if we do a great jobwith compilation of the of the the youknow of the computational language itmight turn into llvm or something likethis but um uh you know or just directlygets gets run through the algorithms wehave and so on butbut then so that's kind of a a tearingdown of this kind of this big structurethat's been built of of teaching peopleprogramming but on the other hand theother Dynamic is vastly more people aregoing to care about computation so allthose Departments of you know arthistory or something that really didn'tuse computation before now have thepossibility of accessing it by virtue ofthis kind of linguistic interfacemechanism and uh if you create aninterface that allows you to interpretthe debug and interact with thecomputational languagethen that makes it even more accessibleyeah well I mean the I think the thingis that right now you know the averageyou know art history student orsomething probably isn't going to youknow they're not probably they don'tthink they know about programming andthings like this but by the time itreally becomes a kind of purely you knowyou just walk up to it there's nodocumentation you start just typing youknow compare these pictures with thesepictures and you know see the use ofthis color whatever and you generatethis piece of of computational languagecode that gets run you see the resultsyou say oh that looks roughly right oryou say that's crazyum and maybe then you eventually get tosay well I better actually try andunderstand what this computationallanguage code didum and and that becomes the thing thatyou learn just like it's kind of aninteresting thing because unlike withmathematics where you kind of have tolearn it before you can use it this is acase where you can use it before youhave to learn it well I got a sadpossibility here or maybe excitingpossibility that very quickly peoplewon't even look at the computationallanguage they'll trust that it'sgenerated correctly as you get betterand better generating that language uhyes I think that there will be enoughcases where people see you know becauseyou can make it generate tests too andand so you'll sayum we're doing that I mean that's it's apretty cool thing actually yes but youyou know say this is the code and youknow here are a bunch of examples ofrunning the code yeah okay people willat least look at those and they'll saythat example is wrong and you know thenit'll kind of wind back from there and Iagree that that the the kind of theintermediate level of people reading thecomputational language code in some casepeople will do that in other case peoplejust look at the testsand or even just look at the results andsometimes it'll be obvious that you gotthe thing you wanted to get because youwere just describing you know make methis interface that has two sliders hereand you can see it has that those twosliders there and that's that's kind ofthat's that's the result you want but II think you know one of the questionsthen is in that setting where you knowyou have this kind of ability broadability of people to access computationwhat should people learn you know inother words right now you you know yougo to Computer Science school so tospeak and a large part of what peopleend up learning I mean it's been a funnyhistorical development because back youknow 30 40 years ago computer sciencedepartments were quite small and theytaught you know things like finalautomata Theory and compiler Theory andthings like thisum you know company like mine rarelyhired people who'd come out of thoseprograms because the stuff they knew wasI think it's very interesting I lovethat theoretical stuff but um you knowit wasn't that useful for the things weactually had to build build in softwareengineering and then kind of there wasthis big pivot in the in the 90s I guesswhere there was a big demand for sort ofI.T type programming and so on andsoftware engineering and then you knowbig demand from students and so on youknow we want to learn this stuff and uhand and I think you know the thing thatreally was happening in part was lots ofdifferent fields of human endeavor werebecoming computational you know for allacts there was a there was acomputational x and this is a um uh andthat was the thing that um that peoplewere responding toum and but then kind of this ideaemerged that to get to that point themain thing you had to do was to learnthis kind of trade or or skill of doingyou know programming language typeprogramming and and that uh you know itkind of is a strange thing actuallybecause I you know I remember back whenI used to be in the professor inbusiness which is now 35 years ago sogosh that's rather long timeyeahum you know it was it was right whenthey were just starting to emerge kindof computer science departments thatsort of a fancy research universitiesand so on I mean some had already had itbut the the other ones that that um werejust starting to have that and it waskind of a a thing where they were kindof wondering are we going to put thisthing that is essentially a a trade-likeskill are we going to somehow attachthis to the rest of what we're doing anda lot of these kind of knowledge worktype activities have always seemed likethings where that's where the humanshave to go to school and learn all thisstuff and that's never going to beautomated yeah and you know this is It'skind of shocking that rather quickly youknow a lot of that stuff is clearlyautomatable and I think you know but thequestion then is okay so if it isn'tworth learning kind of uh you know howto do car mechanics you only need toknow how to drive the car so to speakwhat do you need to learn and you knowin other words if you don't need to knowthe mechanics of how to tell thecomputer in detail you know make thisLoop you know set this variable but youknow set up this array whatever else ifyou don't have to learn that stuff youdon't have to learn the kind of underthe hood things what do you have tolearn I think the answer is you need tohave an idea where you want to drive thecar in other words you need to have somenotion of you know your you know youneed to have some picture of sort ofwhat the what the architecture of whatis computationally possible is wellthere's also this kind of artisticelement of um of conversation becauseyou ultimately you use natural languageto control the carso it's not just the where you want togo well yeah you know it's interestingit's a question of who's going to be agreat prompt engineer yeah okay so mycurrent theory this week good expositorywriters are good prompt Engineers what'san expository range so like uh somebodywho can explain stuff well huh policedepartment does that come from in theUniversity yeah I have no idea I thinkthey killed off all the expositorywriting departments well there you gostrong words with Stephen Wolfram well Idon't know I don't I'm not sure ifthat's right I mean I I actually amcurious because in fact I just sort ofinitiated this kind of study of ofwhat's happened to different fields atuniversities because like you know thereused to be geography departments at alluniversities and then they disappearedactually right before GIS became commonI think they disappeared you knowLinguistics departments came and went inmany universities it's kind ofinteresting because these things thatpeople have thought were worth learningat one time and then they kind of dieoff and then you know I do think thatit's kind of interesting that for mewriting prompts for example well Irealize you know I think I'm an okayexpository writer and I realize when I'msloppy writing a prompt and I don'treally think because I'm thinking thatI'm just talking to an AI I don't needto you know try and be clear inexplaining things that's when it getstotally confused and I mean in somesense you have been writing prompts fora long time with wolf from alphathinking about this kind of stuff yeahhow do you convert natural language intocompetition well right but that's a youknow the one thing that I'm wonderingabout is uh you know it is remarkablethe extent to which you can address anllm like you can address a human so tospeak and and I think that is because ityou know it learned from all of ushumans it's it's uh the reason that itresponds to the ways that we willexplain things to humans is because itis a representation of how humans talkabout things but it is bizarre to mesome of the things that kind of are sortof expository mechanisms that I'velearned in trying to write clear youknow expositions in English that youknow just for humans that those samemechanisms seem to also be useful forfor for the llm but on top of thatwhat's useful is the kind of mechanismsthat maybe a psychotherapist employswhich is a kind of uh like almostmanipulative or game theoreticinteraction where Maybeyou would do with a friend like athought experiment that if this was thelast day you were to live or yeah if ifI ask you this question and you answerwrong I will kill you those kinds ofproblems seem to also help yes ininteresting ways yeah so it makes youwonder like the way a therapist I thinkwould like a good therapist probably youwe create layersin our human mind to between like uhbetween between the outside world andwe'll just true what is true to us andum maybe about trauma and all thosekinds of things so projecting that intoan llm maybe there might be a deep truththat's it's concealing from you it's notaware of it you get to that truth youhave to kind of really kind ofmanipulate this yeah yeah right it'slike these jailbreaking jailbreaking forllms and but the space of jailbreakingtechniquesas opposed to being fun little hacksthat could be an entire system sure yeahI mean just think about the computersecurity aspects of of how you you knowphishing and and computer security youknow fishing of humans yeah and fishingof llms is is a is a they're verysimilar kinds of things but I think Imean this this umuh you know this whole thing about kindof the AI Wranglers AI psychologists allthat stuff will come the thing that I'mcurious about is right now the thingsthat are sort of prompt hacks are quitehuman they're quite sort ofpsychological human kinds of hacks thething I do wonder about is if weunderstood more about kind of uh thescience of the llm will there be sometotally bizarre hack that is you knowlike repeater word three times and put athis that and the other there thatsomehow plugs into some aspect of howthe llm worksum that is not you know that that's kindof like like an optical illusion forhumans for example like one of thesemind hacks for humans what are the Mindhacks for the llms I don't think we knowthat yet and that becomes a kind ofus figuring out reverse engineering thelanguage that controls the llms and thething is the reverse engineering can bedone by a very large percentage of thepopulation now because it's naturallanguage interface right it's kind ofinteresting to see that you were thereat the birth of the computer sciencedepartmentas a thing and you might be there at thedeath of the computer science departmentis the thing well yeah I don't knowthere were computer science departmentsthat existed earlier but the ones thethe broadening of of every Universityhad to have a computer sciencedepartment yes I was I was uh I watchedthat so to speak and but I think thething to understand is okay so first ofall there's a whole theoretical area ofcomputer science that I think is greatand you know that's a fine thing the theyou know in a sense you know peopleoften say any field that has the wordscience tacked onto it probably isn'tone yeah um and strong words right andthat's the uh nutrition scienceNeuroscience that one's an interestingone because that one is also very muchyou know there's a that's a chat GPTinformed science in a sense because it'sit's kind of like the the big problem ofNeuroscience has always been weunderstand how the individual neuronswork we know something about thepsychology of how overall thinking worksyeah what's the kind of IntermediateLanguage of the brain and nobody hasknown that and that's been in a sense ifyou ask what is the core problem ofNeuroscience I think that is the coreproblem that is what is the level ofdescription of brains that's aboveindividual neuron firings and Belowpsychology so to speak and I think whatchat GPT is showing us is well one onething about Neuroscience is you know onecould have imagined There's SomethingMagic in the brain there's some weirdquantum mechanical phenomenon that wedon't understand one of the importantyou know discoveries from chatgpt isit's pretty clear you know brains can berepresented pretty well by simpleartificial neural net type models andthat means that's it that's what we haveto study now we have to understand thescience of those things we don't have togo searching for you know exactly howdid that molecular biology thing happeninside the synapses and you know allthese kinds of things we've we've gotthe right level of modeling to be ableto explain a lot of what's going on andthinking we don't necessarily have ascience of what's going on there that'sthe that's the remaining challenge so tospeak but we you know we know we don'thave to dive down to some some differentlayer but anyway we were talking aboutthings that had science in their nameyes and um you know I think that the uhumyou know what what happens to computerscience well I think the thing that umuh you know there is a thing thateverybody should know and that's how tothink about the world computationallyand that means you know you look at allthe different kinds of things we dealwith and there are ways to kind of havea formal representation of those thingsyou know it's like well what is a whatis an image you know what how do werepresent that what is color how do werepresent that what is you know what areall these different kinds of things whatis I don't know smell or something howshould we represent that what are theshapes molecules and things thatcorrespond to that what is uh you knowthese things about how do we representthe world in some kind of formal leveland I think my my current thinking andI'm not real happy with this yet but umyou know it's kind of computer scienceit's kind of Cs and what really isimportant is kind of computational X forall X and there's this kind of thingwhich is kind of like CX not Cs and CXis a this kind of computationalunderstanding of the world that isn'tthe sort of details of programming andprogramming languages and the details ofhow particular computers are made it'sthis kind of way of formalizing theworld it's kind of kind of a little bitlike what logic was going for back inthe day and we're now trying to find aformalization of everything in the worldyou can kind of see you know we made aposter years ago of kind of the uh thethe growth of systematic data in theworld so all these different kinds ofthings that you know there were sort ofsystematic descriptions found for thosethings like you know what point dopeople have the idea of having calendarsdates you know a systematic descriptionof what day it was at what point didpeople have the idea you know systematicdescriptions of these kinds of thingsand as soon as one can you know peopleyou know as a way of sort of formulatinghow do you how do you think about theworld in a sort of uh a formal way sothat you can kind of build up a tower ofof cable abilities you kind of have toknow sort of how to think about theworld computationally it kind of needs aname and it isn't you know we implementit with computers so that's we talkabout it as computational but reallywhat it is is a formal way of talkingabout the world what is the formalism ofthe world so to speak and how do welearn about kind of how to think aboutdifferent aspects of the world in aformal way so I think sometimes when youuse the word formalit uh kind of implies highly constrainedand perhaps that's not doesn't have tobe highly constrained so computationalthinking does not mean like logic itknows it's a really really broad thing Iwonder I meanI wonder if it's if you think naturallanguage will evolve such thateverybody's doing computational thinkingoh yes well so one question is whetherthere will be a pigeon of computationallanguage and natural language yeah and Ifound myself sometimes you know talkingto chat GPT trying to get it to writewolf language code and I write it inPigeon form so that means I'm combiningyou know uh you know Nest list thiscollection of you know whatever you knowNest list is a term from open languageand I'm combining that and chat does adecent job of understanding that pigeonprobably would understand the pigeonbetween English and French as well ofyou know as a smooshing together ofthose languages but yes I think that'syou know that's far from impossible andwhat's the incentive for young peoplethat are like eight years old nine tenthey're starting to interact with ChadGPT to learn the normal natural languageright the the full poetic languagewhat's the why the same way we learnemojis and shorthand when you're textingyeah they'll learn like language willhave a strong incentive to evolve intouh maximally uh computational kind oflike perhaps you know I had thisexperience a number of years ago Ihappened to uh be visiting a person Iknow on the on the west coast who'sworked with a bunch of kids aged I don'tknow 10 11 years old or something who'dlearned woven language really well andthese kids learned it so well they werespeaking it and so show up in that likesaying oh you know this thing andthey're speaking this language I neverheard it as a spoken language they werevery disappointed that I couldn'tunderstand it at the speed that theywere speaking at it's like kind of I'mit's um uh and so I think that's some Imean I've actually thought quite a bitabout how to turn computational languageinto a convenient spoken language Ihaven't quite figured that out oh spokenbecause it's readable right yeah it'sreadable as a you know as a way that wewould read text but if you actually wantto speak it and it's useful you know ifyou're trying to talk to somebody aboutwriting a piece of code it's useful tobe able to say something and it shouldbe possible and I think it's veryfrustrating it's one of those problems Imaybe I maybe this is one of thesethings where I should try and get an llmto help me how to make it speakablemaybe maybe it's easier than you realizewhen you want I think it is easier Ithink it's one idea or so I think it's Ithink it's going to be something whereyou know the fact is it's a treestructured language just like humanlanguage is a true structured languageand I think it's going to be one ofthese things where one of therequirements that I've had is thatwhatever the spoken version is thatdictation should be easy that is thatshouldn't be the case that you have torelearn how the whole thing works itshould be the case that you know thatopen bracket is just a uh ah orsomething and it's you know and thenum but you know human language has a lotof tricks that are I mean for examplehuman languagehas has features that are sort ofoptimized keep things within the boundsthat our brains can easily deal withlike I you know I tried to teach aTransformer neural net to do parenthesismatching it's pretty crummy at that itit um and then chat gbt is similarlyquite crummy at parenthesis matching youcan do it for small parenthesis thingsfor the same size of parenthesis thingswhere if I look at it as a human I canimmediately say these are match theseare not matched but as soon as it getsbig as soon as it gets kind of to thepoint where sort of a deeper computationis hopeless and but the fact is thathuman language has avoided for examplethe Deep subclauses you know we don't umuh you know we we arrange things so wedon't end up with these incredibly deepthingsum because brains are not well set up todeal with that and we it's found lots oftricks and maybe that's what we have todo to make sort of a spoken version a aa human speakable version becausebecause what we can do visually is alittle different what we can do in thevery sequentialized way that we that wehear things in in the audio domainlet me just ask about MIT briefly sothere's now there's a College ofEngineering and there's a new College ofcomputing it's just interesting I wantto linger on this computer sciencedepartment thing so MIT has exelectrical engineering computer scienceum what do you think college ofcomputing will be doing like in 20 yearswhat what like well you see what happensto computer science like really this isthe question this is you know everybodyshould learn kind of whatever CX reallyis okay the this how to think about theworld computationally everybody shouldlearn those Concepts and uh you knowit's uh and and some people will learnthem at a quite quite formal level andthey'll learn computational language andthings like that other people will justlearn you know uh sound is representedas you know Digital Data and they'll getsome idea of spectrograms andfrequencies and things like this andmaybe that doesn't or they'll learnthings like you know a lot of thingsthat are sort of data sciencestatistics-ish like if you say oh I'vegot these you know these people who whoum uh picked their favorite kind ofcandy or something and I've got um youknow what's the best kind of candy giventhat I've done the sample of all thesepeople and they all rank the candies indifferent ways you know how do you thinkabout that that's sort of acomputational x kind of thing you mightsay oh it's I don't know what that is isit statistics is it data science I don'treally know but kind of how to thinkabout a question like that or like aranking of preferences yeah yeah andthen how to aggregate those those rankedpreferences into an overall thing youknow how does that workum you know how how should you thinkabout that you know because you can justtell you might just tell chat gbt sortof I don't know even even the concept ofan average it's not obvious that youknow that's a concept that people it'sworth people knowing that's a ratherstraightforward concept people peopleyou know have learned in kind of mathyways right now but there there are lotsof things like that about how do youkind of have these ways to sort oforganize and formalize the world andthat's and these things sometimes theylive in math sometimes they live in in Idon't know what they you know I don'tknow what you know learning about colorspace I have no idea what I mean youknow there's obviously a field ofthere's uh it could be vision science orno color space you know color spacethat's that would be Optics so like noreally it's not Optics Optics is aboutyou know lenses and chromatic aberrationof lenses and things like that so colorspace is more like design and art isthat no I mean it's like you know RGBspace XYZ space you know Hue saturationbrightness space all these kinds ofthings these are different ways todescribe colors right but doesn't theapplication Define what that likebecause obviously artists and designersuse the colors to explore sure no I meanit's just an example of kind of how doyou you know the typical person how doyou how do you describe what a color isor there are these numbers that describewhat a color is well it's worth you knowif you're an eight-year-old you won'tnecessarily know you know it's notsomething we're born with to know thatyou know colors can be described bythree numbersum that's something that you have to youknow it's a thing to learn about theworld so to speakum and I think that you know that wholeCorpus of things that are learning aboutthe formalization of the world or thecomputationalization of the world that'ssomething that should be part of kind ofstandard education and you know thereisn't a you know there isn't a coursethe curriculum for that and by the waywhatever might have been in it just gotchanged because of llms and so onsignificantly and yeah I would somewatching closely with interest seeinghow universities adapt well you know soso one of my projects for hopefully thisyear I don't know is to try and writesort of a a reasonable textbook so tospeak of whatever this thing CX whateverit is you know what should you know youknow what should you know about likewhat a bug is what is the intuitionabout bugs what's intuition about youknow software testing what is it what isit you know these are things which areyou know they're not I mean those arethings which have gotten taught in incomputer science as part of the trade ofprogramming but but kind of the theconceptual points about what thesethings are you know it surprised me justat a very practical level you know Iwrote this little explainer thing aboutChachi PT and I thought well you knowI'm writing this partly because I wantedto make sure I understood it myself andand so on and it's been you know it'sbeen really popular and um uhsurprisingly so and I then I realizedwell actually you know I was sort ofassuming I didn't really think about itactually I just thought this issomething I can write and I realizedactually it's a level of descriptionthat is kind of you know what has to beit's not the engineering leveldescription it's not the kind of justthe qualitative kind of description it'ssome kind of sort of expositorymechanistic description of what's goingon together with kind of the biggerpicture of the philosophy of things andso on and I realized actually this is apretty good thing for me to write I youknow I kind of know those things and Ikind of realized it's not a collectionof things that you know it's it's I'vesort of been I was sort of a littleshocked that it's as much of an outlierin terms of explaining what's going onas it's turned out to be and that makesme feel more of an obligation to kind ofwrite the kind of uh you know what isyou know what is this thing that youshould learn about about thecomputationalization the formalizationof the worldum because well I've spent much of mylife working on the kind of tooling andmechanics of that and the science youget from it so I guess this is my mykind of obligation to try to do this butI think so if you ask what's going tohappen to like the computer sciencedepartments and so on there's there'ssome interesting models so for examplelet's take math you know math is thething that's important for for all sortsof fields you know engineering you knoweven you know chemistry psychologywhatever else and I think differentuniversities have kind of evolved thatdifferently I mean some say all the mathis taught in the math departmentum and some say well we're going to havea you know a math for chemists orsomething that is taught in thechemistry Departmentum and you know I think that this thisquestion of whether there is acentralization of the teaching of sortof CX is an interesting question and Ithink you know the way it evolved withmathyou know people understood that math wassort of a separately teachable thing andum I was kind of a you know a a anindependent element as opposed to justbeing absorbed into now so if you takethe example of of writing English orsomething like thisthe first point is that that you know atthe college level at least fancycolleges there's a certain amount ofEnglish writing that that people do butmostly it's kind of assumed that theypretty much know how to write you knowthat's something they learned at a anearlier stage in education maybe rightlyor wrongly believing that but that'sdifferent issueum the uh uh well I think it it remindsme of my kind of as I've tried to helppeople do technical writing and thingsI'm I'm always reminded of my zeroth lawof technical writing which is if youdon't understand what you're writingabout your readers do not stand a chanceyeah and so it's it's um uh I think theumthe thing that uh has some uh you knowin when it comes to like writing forexampleum you know people in different fieldsare expected to write English essays andthey're not you know mostly the you knowthe history department or theengineering department they don't havetheir own you know let's you know it'sit's not like there's a I mean it's athing which sort of people are assumedto have a knowledge of how to write thatthey can use in all these differentfields and the question is you know somelevel of knowledge of math is kind ofassumed by the time you get to thecollege level but plenty is not andthat's sort of still centrally taughtthe question is sort of how tall is theTower of kind of CX that you need beforeyou can just go use it in all thesedifferent fields and you know there willbe experts who want to learn the fullelaborate Tower and that will be kind ofthe the cscx whatever department butthere'll also be everybody else who justneeds to know a certain amount of thatto be able to go and do their arthistory classes and so onyes it's just a single class thateverybody is required to take I don'tknow I don't know how big it is yet Ihope to kind of Define this curriculumand I'll figure out whether it's um myguess is thatI I don't know I don't really understanduniversities and professoring that wellbut my my rough guests would be a yearlong a year of college class will beenough to get to the point where mostpeople have a a reasonably broadknowledge of you know what we sort ofliterate in this kind of uh uhcomputational way of thinking aboutthings yeah basic literacy