Things Glen Found Interesting, Volume 384

On Fri­days I share articles/resources about broad cul­tur­al, soci­etal and the­o­log­i­cal issues. Be sure to see the expla­na­tion and dis­claimers at the bot­tom. I wel­come your sug­ges­tions. If you read some­thing fas­ci­nat­ing please pass it my way.

This is vol­ume 384, which is 8!! (8 dou­ble fac­to­r­i­al). Dou­ble fac­to­r­i­al is a con­cept I learned today. Instead of mul­ti­ply­ing 8 · 7 · 6 · 5 · 4 · 3 · 2 · 1 you instead skip down by twos, so 8 · 6 · 4 · 2 = 384.

Things Glen Found Interesting

1. My AI Safe­ty Lec­ture for UT Effec­tive Altru­ism (Scott Aaron­son, per­son­al blog): “If you had asked any­one in the 60s or 70s, they would have said, well clear­ly first robots will replace humans for man­u­al labor, and then they’ll replace humans for intel­lec­tu­al things like math and sci­ence, and final­ly they might reach the pin­na­cles of human cre­ativ­i­ty like art and poet­ry and music.The truth has turned out to be the exact oppo­site. I don’t think any­one pre­dict­ed that. GPT, I think, is already a pret­ty good poet. DALL‑E is already a pret­ty good artist. They’re still strug­gling with some high school and col­lege-lev­el math but they’re get­ting there. It’s easy to imag­ine that maybe in five years, peo­ple like me will be using these things as research assistants—at the very least, to prove the lem­mas in our papers. That seems extreme­ly plau­si­ble.”
   • Rec­om­mend­ed by a stu­dent.
2. How an Unortho­dox Schol­ar Uses Tech­nol­o­gy to Expose Bib­li­cal Forg­eries (Chanan Tigay, Smith­son­ian Mag­a­zine): “After­ward, the ama­teur archae­ol­o­gist, who would become an emi­nent schol­ar and a mem­ber of the Insti­tut de France, tried to nego­ti­ate with the Bedouin to acquire the stone, but his inter­est, cou­pled with offers from oth­er inter­na­tion­al bid­ders, fur­ther irked the tribes­men; they built a bon­fire around the stone and repeat­ed­ly doused it with cold water until it broke apart. Then they scat­tered the pieces.”
   • Inter­est­ing through­out. Rec­om­mend­ed by a stu­dent.
3. GPT Takes the Bar Exam (Michael Bom­mar­i­to II & Daniel Mar­tin Katz, arX­iv): “For best prompt and para­me­ters, GPT‑3.5 achieves a head­line cor­rect rate of 50.3% on a com­plete NCBE MBE prac­tice exam, sig­nif­i­cant­ly in excess of the 25% base­line guess­ing rate, and per­forms at a pass­ing rate for both Evi­dence and Torts. GPT‑3.5’s rank­ing of respons­es is also high­ly-cor­re­lat­ed with cor­rect­ness; its top two and top three choic­es are cor­rect 71% and 88% of the time, respec­tive­ly, indi­cat­ing very strong non-entail­ment per­for­mance. While our abil­i­ty to inter­pret these results is lim­it­ed by nascent sci­en­tif­ic under­stand­ing of LLMs and the pro­pri­etary nature of GPT, we believe that these results strong­ly sug­gest that an LLM will pass the MBE com­po­nent of the Bar Exam in the near future.”
   • Source code: https://github.com/mjbommar/gpt-takes-the-bar-exam
   • One of the authors (Katz) is a law prof at Illi­nois Tech, and the oth­er (Bom­mar­i­to) is a tech dude as well as an assis­tant law prof (Michi­gan State).
   • Relat­ed: “Extra­or­di­nary new paper from Google on med­i­cine & AI: When Google tuned a AI chat­bot to answer com­mon med­ical ques­tions, doc­tors judged 92.6% of its answers right … com­pared to 92.9% of answers giv­en by oth­er doc­tors.” Source: https://twitter.com/emollick/status/1610261628607512576
4. How the algo­rithm tipped the bal­ance in Ukraine (David Ignatius, Wash­ing­ton Post): “The pow­er of advanced algo­rith­mic war­fare sys­tems is now so great that it equates to hav­ing tac­ti­cal nuclear weapons against an adver­sary with only con­ven­tion­al ones,” explains Alex Karp, chief exec­u­tive of Palan­tir, in an email mes­sage. “The gen­er­al pub­lic tends to under­es­ti­mate this. Our adver­saries no longer do.”
   • Fol­low-up: A ‘good’ war gave the algo­rithm its open­ing, but dan­gers lurk (David Ignatius, Wash­ing­ton Post): “For the Army and oth­er ser­vices, the impe­tus for this tech­nol­o­gy push isn’t just the Russ­ian inva­sion of Ukraine, but the loom­ing chal­lenge from Chi­na — America’s only real peer com­peti­tor in tech­nol­o­gy.”
5. The Con­ser­v­a­tive Who Wants to Bring Down the Supreme Court (Jean­nie Suk Ger­son, The New York­er): “One of Mitchell’s close friends from law school is a female lawyer who is mar­ried to a woman. She recent­ly told her teen-age daugh­ter that, if their fam­i­ly ever need­ed some­one to donate an organ, she knew they could call on him. ‘But, at the same time, his views, the results of his views, and his pol­i­tics felt not nice, to put it mild­ly,’ she said. ‘I always assumed that, since Jonathan is such a good per­son, that when he aged and knew more peo­ple, his views would evolve. I real­ly have trou­ble rec­on­cil­ing these two parts of him, giv­en my pol­i­tics and my view of the world, because I just find him to be such a kind, lov­ing per­son.’ But Mitchell doesn’t strike her as ‘a true believ­er who will mar­shal his argu­ments to jus­ti­fy the out­come,’ she said. ‘I think he actu­al­ly believes these legal argu­ments.’ ”

Less Serious Things Which Also Interested/Amused Glen

• Math and Real Life (SMBC)
• Men’s Bod­ies Are Way More Dis­gust­ing Than Wom­en’s (Dry Bar Com­e­dy, YouTube): four min­utes
• Miran­da Hart Can’t Believe Jack White­hal­l’s Cat Paint­ing Abil­i­ty! (Would I Like To You?, YouTube): five min­utes

Things Glen Found Interesting A While Ago

Every week I’ll high­light an old­er link still worth your con­sid­er­a­tion. This week we have The Coro­n­avirus and the Right’s Sci­en­tif­ic Coun­ter­rev­o­lu­tion (Ari Schul­man, The New Repub­lic): “That so many views tut-tut­ted as the irra­tional defi­ance of expert con­sen­sus actu­al­ly became the expert con­sen­sus in the span of just a few weeks vivid­ly sug­gests that we need to reex­am­ine just how our cul­ture talks about exper­tise. The prob­lem is not main­ly that the experts were wrong—that is to be expect­ed. It is, rather, that our lead insti­tu­tions and pub­lic infor­ma­tion out­lets con­tin­u­al­ly treat­ed the assur­ances of experts as neu­tral inter­pre­ta­tions of set­tled sci­ence when they plain­ly were not.” Inter­est­ing through­out and still rel­e­vant. From vol­ume 259

Why Do You Send This Email?

In the time of King David, the tribe of Issachar pro­duced shrewd war­riors “who under­stood the times and knew what Israel should do” (1 Chron 12:32). In a sim­i­lar way, we need to become wise peo­ple whose faith inter­acts with the world. I pray this email gives you greater insight, so that you may con­tin­ue the tra­di­tion of Issachar.

Disclaimer

Chi Alpha is not a par­ti­san orga­ni­za­tion. To para­phrase anoth­er min­is­ter: we are not about the donkey’s agen­da and we are not about the elephant’s agen­da — we are about the Lamb’s agen­da. Hav­ing said that, I read wide­ly (in part because I believe we should aspire to pass the ide­o­log­i­cal Tur­ing test and in part because I do not believe I can fair­ly say “I agree” or “I dis­agree” until I can say “I under­stand”) and may at times share arti­cles that have a strong par­ti­san bias sim­ply because I find the arti­cle stim­u­lat­ing. The upshot: you should not assume I agree with every­thing an author says in an arti­cle I men­tion, much less things the author has said in oth­er arti­cles (although if I strong­ly dis­agree with some­thing in the arti­cle I’ll usu­al­ly men­tion it). And to the extent you can dis­cern my opin­ions, please under­stand that they are my own and not nec­es­sar­i­ly those of Chi Alpha or any oth­er orga­ni­za­tion I may be per­ceived to rep­re­sent. Also, remem­ber that I’m not report­ing news — I’m giv­ing you a selec­tion of things I found inter­est­ing. There’s a lot hap­pen­ing in the world that’s not mak­ing an appear­ance here because I haven’t found stim­u­lat­ing arti­cles writ­ten about it. If this was for­ward­ed to you and you want to receive future emails, sign up here. You can also view the archives.