The Swyx Mixtape

Listen to the InfoQ podcast: https://soundcloud.com/infoq-channel/interview-shawn-swyx-wangBlogpost: https://dx.tips/the-end-of-localhostIn this episode, Shawn Wang (swyx), head of developer experience at Airbyte, sat down with InfoQ podcast co-host Daniel Bryant and discussed the rise of remote development environments. Topics covered included, whether remote development experiences are good enough to see the death of local(host) development, what a wishlist might look like for the ultimate developer experience, and how cloud native organizations are currently developing software.Read a transcript of this interview: bit.ly/3R3OEcDSubscribe to our newsletters:- The InfoQ weekly newsletter: bit.ly/24x3IVq- The Software Architects’ Newsletter [monthly]: www.infoq.com/software-architects-newsletter/Upcoming Events:QCon San Francisco:qconsf.com/- Oct 24-28, 2022- Oct 2-6, 2023QCon Plus online:plus.qconferences.com/- Nov 29 - Dec 9, 2022QCon Londonqconlondon.com/- March 26-31, 2023Follow InfoQ:- Twitter: twitter.com/InfoQ- LinkedIn: www.linkedin.com/company/infoq- Facebook: bit.ly/2jmlyG8- Instagram: www.instagram.com/infoqdotcom/- Youtube: www.youtube.com/infoqDXSWYXLocalHostRemotedevelopment

Make you feel my love: https://www.youtube.com/watch?v=Qxsd7rhhFYwAin't No Mountain: https://www.youtube.com/watch?v=yB5ITxz9CK0Easy on me: https://www.youtube.com/watch?v=lb4f5IPvwXU

Watch: https://www.youtube.com/watch?v=UQkeRxeh34ITranscripteric schmidt is a business leader andsoftware engineer that served asgoogle's chief executive officer from2001 to 2011.under his leadership google grew from anearly silicon valley startup to arguablythe most important technology company onthe planetschmidt is currently co-founder ofschmidt futures and sits on the board ofmany public and private institutionshe is still involved with technologyconsults with the us department ofdefensealso talks about ai in his latest bookthe age of ai and our human futurewritten alongside former u.s secretaryof state henry kissinger and computerscientist daniel huddenlockerschmidt was a guest at the milken globalconference and here he anticipates someof the ai innovations that we willcertainly see in five yearshe also predicts what we might see in 20yearshere are the detailsrecently in the last couple of yearsthere have beenextraordinary gains so for example ateam at google and at the baker labseparatelyfigured out a way to actually understandif you take dna what proteins aregenerated and what their structure isthat's an extraordinary achievement inmy opinion worth a nobel prizethere are drugs being designed now thatcould not be possiblydesigned by humansin any way because of their complexitythere's evidence that ai can be used inbiology ai is mapped to biology the waymath is to physics in other wordsbiology is so complicated that ai willbe used to interpret biology and predictits outcomeover and over again ai will arrive inyour lifeanother example is the hottest area inmy industry right now are large languagemodels uh recently a set of startupshave been funded between 100 and abillion 100 million and a billiondollarsthey have no current product or revenueplans umthe the belief of the power of thistechnology these large language modelsare interesting because you suck all theinformation inlike you read all the web whichcomputers can do but we can't and thenthey discover things they appear todiscover a structure of language and anexample of recent google product lastweek can actually translate from onecomputer language to another and wedidn't give it any examples of one totranslate to the other it discovered astructure and it can predict itthese are the beginning of generalintelligencethe the current um excitement stems froma technology called transformers thatwas invented three or four years ago andwhat transformers do is it can predictthe next word after a set of words so ifyou give it a sentence it can predictwhat the word will be and it's doneusing a complicated mathematicaltechnique it turns out predicting thenext word is mathematically the samething as predicting the next sound thenext video the next imageall of that and so you have aunification a multi multi-modalunification of video text and speech sothese systems sound and look likethey're intelligenta good example is gpt3 which came outlast yearwhich kicked the current revolution offyou asked itdo you think like a human and it says noi do notbecause i am a large language model andyou are a i think a human who has beentaught to think in this waynow is thatit thinking about you or is it patternmatching we can't tell and the truth isand i'm as part of philanthropic worki'm funding projects to try tounderstand this we don't actuallyunderstand why this works we don'tmathematically understand why it worksand we also don't understand its failuremodesso you wouldn't want to use this as areplacement for something that's livecritical because we can't say when itfails when does it just crashthe current large language models forexample have trouble with the notion ofgravity so if you say to them i moved iti moved the object from here to here andthen i put it up here and i put it downthere and so forth now everyone justfollowed what i did the large languagemodel gets confused because it doesn'tunderstand gravity so the computerscientists say we're going to now addconceptsrightso with concepts and then with planningmaybe you get to the point where itlooks like a human-like intelligencewhich has all sorts of issuesif i were 24 today this is exactly whati'd be working on this is where thehardest and most challenging computerscience systems problems are with thegreatest payoffnow remember that the system can predictpatternsand if you can predict a pattern you canalso generate an artifact there's aduality in these systems where they cangenerate thingsso part of the issues that we face nowis that these systems can generatespeech i'll give you an examplewithin five yearsthe following will be trueyou'll be able to take a systemum take one of these language modelswhich would be infinitely expensive tomake but you didn't pay for itit shows up in your doorstep and it finetunes the technical term is literallyfine-tuning it you fine-tune it to youwho are you what do you care about itsort of watches you and learns from youit learns your voiceright all of a sudden it can generatevideos with you in itnow you could think of this as a secondai rightnow the interesting thing is imaginefive years from now i install this thingand i use it for a few years andeventually we all die unfortunately wellit lives onrightas a pretty good impersonation of meand what happens when i'm dead and it'sstill learning is that meis that an artifact of me or is it justa stupid artifact of history that you'llkeep in a box and some future will sayoh eric was so stupid back then but it'sentertaining to watch him right becausehe didn't keep learning we don't know wehave no way of discussing these thingsthis stuff is incredibly powerful itwill be the basis of enormous gains inhuman healthlanguage translation communicationsummary and educationall the things that milken representswill be affected in an almost alwayspositive way having said that there'sterrifying consequences as well so thefirst question has to do with jobsdoes this fundamentally mean there aremore jobs or less jobs i spent my wholelife people saying computers willreplace humans humans won't haveanything to do so far that narrative hasbeen false notice that there's a hugesurplus of jobs and not people to fillit certainly in the united states thesecond one has to do with nationalsecurity something i've worked on foralmost a decade nowand in our in the kissinger book we talka lot about thiswhat happens when the...

Listen to Cautionary Tales: https://www.listennotes.com/podcasts/cautionary-tales/cautionary-conversation-the-FW0bxClD_ij/

Listen to Ted Talks Daily: https://www.listennotes.com/podcasts/ted-talks-daily/the-marvels-and-mysteries-JkbNAa9DMYl/ (or video: https://www.ted.com/talks/heidi_hammel_and_nadia_drake_the_marvels_and_mysteries_revealed_by_the_james_webb_space_telescope?language=en)more with Heidi: https://www.youtube.com/watch?v=oMbeCKh-9v4

Listen to Revisionist History and Inside Voice: https://www.listennotes.com/podcasts/revisionist-history/from-inside-voice-lake-bell-8XVyoz7eT5A/

LIsten on YouTube: https://www.youtube.com/watch?v=O1puRnV6gwE

Listen to the Chopping Block:Nov 9 https://www.listennotes.com/podcasts/unchained/the-chopping-block-ftx-the-3ki-p42chnb/Nov 16 https://www.listennotes.com/podcasts/unchained/the-chopping-block-why-rlhFySZAtMB/

- Listen to Data Eng Podcast: https://www.dataengineeringpodcast.com/data-ecosystem-year-in-review-2021-episode-251/- How dbt Fails: https://benn.substack.com/p/how-dbt-fails

Watch/listen to Analytics Everywhere: https://www.youtube.com/watch?v=vmPvZ_YRSgsGitlab data guide: https://about.gitlab.com/handbook/business-technology/data-team/platform/dbt-guide/ and https://about.gitlab.com/handbook/business-technology/data-team/Our previous episode on dbt: https://mixtape.swyx.io/episodes/wtf-is-dbt-drew-banin