Alex Burnap

Assistant Professor of Marketing @Yale. Machine learning for product design and product management.

[YCCI @ Yale SOM @YaleCCI]

When it comes to marketing cars, form matters as much as function. @YaleSOM Prof. Alex Bernap, @MITSloan Prof. John Hauser and @KelloggSchool Prof. Artem Timoshenko discuss how AI is being used to design more appealing cars.

https://t.co/PrLBfGnKeZ

[Eliezer Yudkowsky ⏹️ @ESYudkowsky]

I don't think people realize what a big deal it is that Stanford retrained a LLaMA model, into an instruction-following form, by **cheaply** fine-tuning it on inputs and outputs **from text-davinci-003**.

It means: If you allow any sufficiently wide-ranging access to your AI model, even by paid API, you're giving away your business crown jewels to competitors that can then nearly-clone your model without all the hard work you did to build up your own fine-tuning dataset. If you successfully enforce a restriction against commercializing an imitation trained on your I/O - a legal prospect that's never been tested, at this point - that means the competing checkpoints go up on bittorrent.

I'm not sure I can convey how much this is a brand new idiom of AI as a technology. Let's put it this way:

If you put a lot of work into tweaking the mask of the shoggoth, but then expose your masked shoggoth's API - or possibly just let anyone build up a big-enough database of Qs and As from your shoggoth - then anybody who's brute-forced a *core* *unmasked* shoggoth can gesture to *your* shoggoth and say to *their* shoggoth "look like that one", and poof you no longer have a competitive moat.

It's like the thing where if you let an unscrupulous potential competitor get a glimpse of your factory floor, they'll suddenly start producing a similar good - except that they just need a glimpse of the *inputs and outputs* of your factory. Because the kind of good you're producing is a kind of pseudointelligent gloop that gets sculpted; and it costs money and a simple process to produce the gloop, and separately more money and a complicated process to sculpt the gloop; but the raw gloop has enough pseudointelligence that it can stare at other gloop and imitate it.

In other words: The AI companies that make profits will be ones that either have a competitive moat not based on the capabilities of their model, OR those which don't expose the underlying inputs and outputs of their model to customers, OR can successfully sue any competitor that engages in shoggoth mask cloning.

[Alex Burnap @AlexBurnap]

@benmschmidt @deaneckles Thanks for retweeting @deaneckles @benmschmidt . Curious to hear your thoughts on competitive market implications of this on GOOG/FB/MSFT? Seems like if this is an “arms race,” setting these closed-source precedents may lead to other peer-tech firms following suit.

[Han Zhao @hanzhao_ml]

Excited to announce that we are organizing Machine Learning for Consumers and Markets Workshop (MLCM) at KDD 2021! Please consider submitting your great work on connecting AI & Business to our workshop! The submission deadline is May 10th, 2021.
Website: https://t.co/BvLwfvquSv

[Kosuke Uetake @Chanman_ECON]

Very proud to see that my former student Minkyung (@MinkyungKim330)'s dissertation-based paper, published in JMR 2019 received 2021 Don Lehmann Award https://t.co/hnw6DqNFqI

[Dean Eckles @deaneckles]

Journals often don't follow a "Pottery Barn rule" (@hardsci) of publishing (eg) novel data replications of papers they have published. But many at least have a venue for corrective notes (eg Letters at PNAS).

I guess some econ journals don't have that: https://t.co/bl9Ej9dQH4

[Alex Burnap @AlexBurnap]

@lil_dill ...

[Alex Burnap @AlexBurnap]

Awesome wrap up to a semester filled with amazing speakers including @lennysan and @jackiebo!

[Alex Burnap @AlexBurnap]

@Lewis_Lin thank you for speaking in Product Management class @YaleSOM.

[Alex Burnap @AlexBurnap]

@Chanman_ECON Dunno about a paper but invite-only products have been a thing for awhile— Gmail, now Clubhouse

[Alex Burnap @AlexBurnap]

@jondr44 Twitter bubbles

[Alex Burnap @AlexBurnap]

@deaneckles what do you think of this article? I feel like you have nuanced perspective here.

[Zachary Lipton @zacharylipton]

“Today it is possible to find tenured professors of management who have never set foot inside a real business, except as customers.”

Possible? 🤣

Interestingly, far more CS faculty work for & with companies, & many more understand management.

https://t.co/aWBlWDmTVp

[Alex Burnap @AlexBurnap]

@zacharylipton Consulting itself ranges from a couple week PPT presentation (a la MBB mgmt consulting) to more in-depth engagements. Somewhere in between what I’ve seen anecdotally with CS profs. That combined w/ time opportunity is the variance I meant.

[Alex Burnap @AlexBurnap]

@zacharylipton Well this is only anecdotal, but the incentive structure of CS profs is write grants and become a manager of grad students. B-school profs don’t have that structure, and have more opportunity to consult.

[Alex Burnap @AlexBurnap]

@zacharylipton (2) the executive mbas tend to impart their experience and real problems on faculty.

[Alex Burnap @AlexBurnap]

@zacharylipton Totally agree with the article in many ways. There are however a few transfer mechanism of “real problems” from companies that business schools tend to do better though is (1) more varied consulting by profs on average, rather than CS profs that tend to take extended sabbaticals

[Bernhard Schölkopf @bschoelkopf]

This may be a good starting point for people thinking about how causality can help tackle some of the open problems of machine learning. It discusses core assumptions such as "independent causal mechanisms" and "sparse mechanism shift".

[Alex Burnap @AlexBurnap]

@johncutlefish @Amplitude_HQ Please do! Those of us teaching PM have to use simulated or often limited case study data due to NDAs. If you end up going with that idea please pm me and I’d love to chat more.

[Gergely Orosz @GergelyOrosz]

“We don’t write PRDs (Product Requirement Documents) as that’s too waterfall. We’re more agile than that.”

My experience tells otherwise: you’ll prob lose a lot of time later. A good PRD answers a few, but critical questions. Here’s how short it can be at Airbnb (via @lennysan )

[Alex Burnap @AlexBurnap]

@faderp @Wharton @Penn @KarynAmira First time I’ve heard of brisket in marketing. @KarynAmira we talking brisket for 14 hours at 225F?