You've likely heard the quote that "every company is a technology company", mostly used in the context of companies needing to be doing everything they can to leverage and exploit new technology, or they'll be left in the dust by more proactive competitors.
With that in mind, a key quote for me in the article was this:
>To help spur internal disruption, the company keeps tabs on 2,000 technology ventures, using about 100 in pilot programs that will eventually join the firm’s growing ecosystem of partners.
I think there will be a role emerging within Enterprises, (sort of like an "insurance broker" who finds the best coverage options among a sea of choices), that will be responsible for surveying new technologies that could potentially help the firm, and running pilot programs to see how they potentially help make the org. more efficient or more competitive.
One could argue that this is really the CTOs role, and it is, but often CTOs are focused on building their own technology, or maintaining existing legacy systems, and not so much on finding and implementing new technology that has not yet hit the main stream. So, at least the focus needs to shift.
Quite a few banks have something called "Innovation Labs" or similar whose job is to do exactly that: match emerging technology with appropriate internal use-cases. Part need-finder and part technologist. Capital One Labs, for instance, was crucial in helping the startup where I work land its first F500 deal. I'd imagine that "innovation labs"-type units are probably emerging at enterprises outside of finance as well.
> I think there will be a role emerging within Enterprises, (sort of like an "insurance broker" who finds the best coverage options among a sea of choices), that will be responsible for surveying new technologies that could potentially help the firm, and running pilot programs to see how they potentially help make the org. more efficient or more competitive.
You have just described my role. We're calling ourselves Emerging Technology Specialists for the time being.
As someone who built a major crypto exchange working on a second fintech business, this sounds interesting. Can you explain a bit more about where you work, the types of tech you've worked with and how you see this evolving? Feel free to reach me by email if you'd prefer not to discuss here.
Very cool. I was just thinking about this yesterday. It'd be really interesting to have the roll of researching the upcoming techs and trying to see how they'd integrate. Are you a start up or do you work for a larger corporation?
It depends on whose definition you use with respect to the CIO/CTO role. In my experience, CIOs are typically more focused on the internal IT infrastructure, keeping it secure, running smoothly, etc., (i.e. an more inward facing almost operational role), whereas a CTO should be outward facing looking and building technological advantage for the firm (i.e. a more outward facing commercial role).
However, I've seen it argued both ways. I'd be interested to hear what others think on the differences in the CIO and CTO roles.
Why would the Chief Investment Officer at a bank be involved in which technologies to adopt? Or did you mean Chief Information Officer or Chief Innovation Officer (which Wikipedia says exists, but I had never heard of)?
Meanwhile I'm going through a mortgage refi and the mix of old and new technology is comical. I was hoping things had advanced since I last went through this a decade ago but not so much. This is through eRates Mortgage and you might think the "e" stands for electronic, but I'm not so sure.
So far I've had to:
- Fill out an online credit application form, which among other things ask me what year I started my last job and converted that to "years and months" but was off by one since it appeared to be doing math as if the current year is 2016. That was at one particular web site.
- On another web site, I was then able to e-sign about a dozen documents. However, about half a dozen documents I had to print out, sign, scan, then upload to a third web site.
- One of the forms had an error, so I received an email from a mortgage officer that read "You've received an encrypted message from XXX@financeofamerica.com. To view your message Save and open the attachment (message.html), and follow the instructions. Sign in using the following email address: me@example.org". This comically required me to register with live.com so that I could read the actual email and reply with a corrected version of the form via an outlook web interface.
- Yet another mortgage officer contacted me for additional forms, these were sent as PDF attachments that I had to print, sign, scan, and email back (over regular email this time) as further attachments.
- There is no online site I can use to track the progress of things. I just have to call or email periodically.
- The forms as you might imagine are a mix of ancient government mandated forms and sundry other requests that are from the loan company's underwriting department.
On a semi-related note, I also just rolled over a 401(k) which was also unable to be done electronically. The old 401(k) institution overnighted me a check which I had to combine with a form I printed out from the new 401(k) institution, then I had to overnight those to the new institution. 3+ days of lost market gains.
The mortgage documentation problem is on my list of things to work on starting Q3 2017. I'd love to hear more about what's pissing you off -- may I email you?
Had a company that was solving sort of this problem a while back (early 2000s). Read in forms (1003 for example), did automated underwriting (hooked in with Fannie Mae), pulled in rate sheets from different Banks to do automatic rate calculations for loan officers. Rate sheets were loaded in daily. There was a rules engine to cover the many complex rules required to determine if a loan was viable for a given 1003 form, etc.
The issue has never been the backend systems at banks -- I've worked with that stuff starting from origination through securitization and on to secondary market trading.
The problem is the customer abuse ("needs list") that is a hallmark of underwriting. There has to be a better way to get all that information from the customer in the vanilla case.
> The issue has never been the backend systems at banks
I find this hard to believe considering the pains we were solving: now quite some time ago. Maybe they have changed.
Excessive burden on mortgage consumers is driven by many aspects.
At it's core is (was if what you say is true) the inability for backend systems to effectively aggregate and verify the necessary context to easily meet regulatory needs.
We were starting to aggregate such data allowing us to automatically verify and then fill out all forms necessary for the mortgage consumer.
And a world where if your contract does not meet with the approval of the computer, you are screwed with no recourse just like you are when Paypal decides to freeze your account because one of their algos detected "fraud"
> when Paypal decides to freeze your account because one of their algos detected "fraud"
Even thinking about the times that happened to my company brings red in front of my eyes. Not many companies are that bad luckily. Google adsense/adwords comes close.
Paypal does naive keyword matching. I know this, because I used the word 'aleph' in the comments field when sending someone money, because that word was in the name of the thing I was buying.
Turns out, there is some terror group with the same word in their name.
Also turns out, PP's "sophisticated" fraud detection will hold a payment for nearly a month, when a human looking at the transaction is going to figure out it is a false-positive from a bullshit system pretty quickly.
Also turns out that apparently terrorists are somehow still capable of functioning, despite the best minds of our generation writing these super-cutting-edge detection algorithms.
I have to imagine that there are members of the set of people who are terrorists and also stupid enough to type "for the greater glory of MyTerrorGroup!" in a PP comment field, but there can't be many. And guess what? This gets everyone thinking about what sort of scanning takes place. So in the interests of catching the two idiot financiers who do this, we're teaching everyone to Be Careful What You Say.
Which, come to think of it, is probably an outcome that some are not unhappy about.
Companies like PayPal aren't doing this because they honestly believe it will catch terrorism. They are doing this because the US government tells them to, and they risk onerous penalties for non-compliance.
"Aleph" is on that list as an alias of Aum Shinrikyo, the cult responsible for the Japanese subway Sarin attack in 1995.
These regulations are somewhat pointless in that the list is public so anyone on the list can easily find out the fact they are on it and use an alias / front company / etc which isn't – so they only catch the truly incompetent terrorist funder/sanctions-buster/etc. But I don't think we can blame companies like PayPal for the existence of government regulations like this–I'm not aware that PayPal, or any other major corporation, has lobbied for these regulations to exist–although if PayPal take a month to remedy a false positive, that is unacceptably slow and it is fair to criticise them for that.
Question is, does the gov specify what actions to take when a word is encountered? Sure scan for them, maybe report them but banning, blocking etc without proper investigation seems more a typical Paypal thing I have come to expect over the years working with them than something the US gov would specify.
If you apply your best efforts to following the US government's advice, but then unintentionally end up doing business with a denied party – they are unlikely to prosecute you, and any prosecution is unlikely to suceed. If you ignore their advice, and then end up doing business with a denied party – even if you never realised they were one – then negative legal consequences are much more likely.
Any US exporter (including overseas subsidiaries of US corporations) is required to obey these laws. Other countries have similar laws which must be obeyed as well. For a large global business, this can get very messy fast, which is why large businesses employ whole teams of people to manage this and use software to help them do it (many ERP packages incorporate this functionality).
And these laws don't say "report the transaction to the government but let it through anyway". The laws say "report the transaction to the government AND deny it". If you do the former instead of the later, you are likely both breaking the law and telling the government that you've done it, which is unlikely to end well for you.
From an end-user perspective, freezing the suspect transaction until an investigation can confirm it is harmless is unpleasant. But, from the company's viewpoint, if they don't take the time to do a through investigation of a suspect transaction, they are exposing themselves to significant legal risks.
(Disclaimer: I speak for myself, and these statements should not be taken as statements of my employer. I am not a lawyer and I don't work in the export control field.)
come on. there are >6b people on earth, with many many fraudsters trying to make money at other people's expense. It's a problem that has to be automated sooner or later.
Computers are also much faster in fraud detection. You use your credit card at some store that you have never been, at an odd hour, and your bank triggers a fraud lock, and declines the transaction. That's efficiency.
When there are billions of transactions occurring every day, of course a small error rate will reflect itself on a large number of false positives.
PS: goog employee, but i appreciate fraud detection a lot (especially when it comes to how easy it is to leak SSN in this country, and how useful it is for obtaining credits)
Edit: fact check. there aren't 6b people using computers obviously.
> That's efficiency. When there are billions of transactions occurring every day, of course a small error rate will reflect itself on a large number of false positives.
Yes, coupled with excellent customer service that sounds like a good plan.
Unfortunately your employer seems to have decided that customer service is something they can't afford.
It's actually much worse than "not known for super good customer service", Google are known for awful customer service and not just for the "free" tier. I appreciate that as your employer Google means something different to you, and you certainly shouldn't be expected to answer on their behalf ... but lets not kid ourselves here
When I write here, i try to write it on my own. Being a google employee doesn't have any bearing on what I think about google products. I don't use many google products, because there are better ones out there. I think that should be the perspective when you read most comments here.
But the fraud algo's + the support are absolute crap. I am known to Google + Paypal, I am logged on as a user, I have credit cards and bank accounts attached, I am with both about since they came online. I made money for both; I am not talking as a consumer, i'm talking as a business owner. Which is not the 6b people on earth; it's far less.
The problem is that companies actually, legally steal money like that; I did not fraud anyone ever and yet I have to provide proof (which I already did 100 times; they have my passport copy, bank statements, where I live, know me for 15+ years etc (and have data on me for that time)) and then still smug (...) canned responses that 'this is the way it is'. But I have no recourse besides legal action which is usually not worth it. So I just have to suck it up. It's theft; I did not do anything. That is my gripe; outside internet, it would not stand. But luckily I left that all behind many years ago; this just brought it back in my head :) Sorry for ranting.
I'm all for automation, but not for bad automation. Google is still party to that in many ways; I am logged into Google / Gmail / Plus / Drive / Photos and do a search; I get a captcha request because I 'might be a bot', then I fill it correctly and it says my computer might be sending automated queries. IT IS NOT and it's just a disgrace to the company, when I am logged on as a very long term paying loyal user, that they cannot simply see 'oh, that's Tycho, let's not annoy him with stuff that he never does/did/will (and he uses Linux, is a programmer, uses Noscript so his laptop is very much not compromised) do and oh yeah he is logged in via 2-factor auth so it's him'. Boggles the mind. Not sure how you can defend it.
If it has happened repeatedly caused that kind of problem you might start looking around for a class action lawsuit. Absent any consiquences they'll keep doing what they do.
At the time (this is a long time ago) we looked into that; there were (maybe are, luckily we left ads behind) many conspiracy theories and angry blog posts from people who had their life wrecked by an algorithm without any way to defend themselves. Nothing came of it and we/I learned that you cannot count on Paypal or Google for primary income; we switched to affiliate sales and traditional online card payments and we never looked back.
I am in ads, and occasionally deal with ad fraud. So i cannot speak what they do for other products. I don't think you should derive any meaning from what I say.
Maybe you can answer a question I had for a long time. Many years ago my company got banned from adsense suspicious traffic. We got some standardised email, we mailed back and got some canned responses but no-one cared of course. I have no clue what it even means because no-one explained (besides links to vague blahblah), so 1) what does that mean? I can guess what it means, but how, if we didn't do it (and I seriously couldn't give a crap anymore at this time, but we had nothing to do with it), how can we be banned for it? Then 2) if Google can detect this traffic and qualify it, why not just delete it from the stats and leave it at that?
Edit; not saying you work there so you know it, but if you do and can say, i'm curious :)
it happ to me. I called them. they said there is nothing you or your buyer did / can do. When i asked PP what the heck is it? they said to me: "Concentrated buyer exposure" I got the buyer on the phone with PP and me. PP said only option is for me to refund buyer then be paid outside PP OR wait for 90 days then i can access my money. It was the first time this ever happ to me, i set up stripe the very next day and am very happy. Not doing PP. Algo's have NO compassion or ANY contextual understanding of why the buyer is paying me ( this case was past invoices, paid in 1 shot like 4 ) guy on the phone , first thing he did was read his PP "script" this conversation can be recorded. What a PITA. He could do "jack" and "shit" to help me. Its my money. How could PP just freeze my money without recourse? WHO reads the FINE print?
I guess this is due to a combination of multiple factors:
- your average of PP transfer amount per transaction is fairly low
- that single buyer / transaction was way over the usual average, possibly in the five-figures range
- the buyer paid PP with a credit card
With the last item, it begins to make sense: the dispution timeframe is 90 days (per FTC, https://www.consumer.ftc.gov/articles/0219-disputing-credit-...). Paypal might be considering a possible dispute in small amounts (maybe 100$) as "chump change" but doesn't want to be on the hook for getting 10k $ back from the seller if the buyer wants to do a dispute.
If I were a one-man-payment broker, I would do exactly the same. No way I'd accept such a liability, no matter how big or reliable your business is - it's not you who is the risk, it's the buyer who is a direct risk to my business because I'm at the interface with the CC issuing bank which will hit me with a negative 10k$ balance and lawyer fees to resolve the matter and get my account balanced.
But, thank the heavens, I live in Germany where we have way more transactions done using SEPA direct transfer (online direct payment, next to impossible to revoke except for technical errors or fraud, and even if it's only possible in 10 days) or SEPA debit, which does have revocations too, but unlike with CCs, the issuer banks usually punish the customer harshly for wrongful revocations.
Probably not any time soon. The issue is that when push comes to shove, a contract is something that gets interpreted by a court. Judges are okay calling in expert witnesses to look at various kinds of evidence, but contract interpretation is a core part of their responsibility and they aren't going to willingly pass that off to computers.
You might see automated arbitration and evolution of contract language to facilitate that, but even given the strong laws regarding the enforceability of arbitration agreements the courts are still are the ultimate backstop and have to be able to read and understand a contract.
Unfortunately, I don't believe that is correct. To the extent of my understanding, the vagueness doctrine applies only in some criminal law circumstances, while contract litigation would generally be within the realm of civil law, so it would be inapplicable. There may, however, be relief available under doctrines like that of unconscionability.
In Chicago JPMorgan is where a lot of software developers go to work for 5-7 years before getting a VP promotion and leaving for a C level role. I've seen dozens of developers follow this exact track. Glad to hear they are doing something interesting during that time.
A VP in a bank is mid-level, an investment bank will have literally thousands of VPs. You could hop from MD (2 levels above VP) to the c-suite perhaps, but there are still hundreds of MDs.
Did a quick glance at linkedin connections who use to work there. Seems a lot of them started their own company or moved to director level roles. Some C level.
My belief is that we won't have (real) self-driving cars until long after we've automated lawyers -- legal work being a much more constrained, simpler problem without too much potential to kill people.
I bet the bank is offloading to "the system" the niche find-and-seek jobs that interns did before.
Something like "find me all the contracts for mortgages in seattle between 1996 and 2004 on fixed rate loan and with guarantees in the same state, and delaware, and _____".
Then bundling all to a senior lawyer and then offloading to the CRM.
Absolutely. If your job is data in -> data out, your job will be taken by a computer. Most of the work lawyers do is just this. Human lawyers will be relegated to client relations and high level strategy. Most documents will be first drafted by a computer, then edited by a human. So law firm partners will still have work, but the need for associates will dry up as computers take over
Definitely not for full automation. The question "do I have a case?" could involve basically anything, and sounds very AI-complete to me.
Of course, discovery and a lot of the repetitive work can be automated much more easily, but that's more like keeping the car in a highway lane, which we can already do for cars.
Attorney here. Frequently do case evaluations for my firm. I prototyped this last year. It was not as hard as you would think if you have some experience in the area of law.
In the early 1990's, JP Morgan developed software tools it packaged and sold as "RiskMetrics." It standardized the way most banks measured and reported their portfolio risk. For example, it was now easy to determine how likely a portfolio would lose, say, 5% of its value. This underlying concept, called Value at Risk, was baked into the RiskMetrics tool and quickly became the default methodology for measuring risk among traders but also regulators, bank management, and external portfolio managers and investors.
The problem was VaR had never really been tested in a real-world meltdown situation. Its models all assumed even under duress, markets acted rationally and traders' behavior was uncoordinated and uncorrelated. In 1998, the spectacular failure of Long-Term Capital Management, was enabled by its reliance on VaR (among other things, like bad trades and enormous leverage). In a crisis, markets were NOT acting rationally, and neither were traders, causing losses much larger than anticipated by VaR. Unfortunately, people tend to repeat mistakes, so in 2007-8, once again, banks all over the world that relied on VaR calculations to tell them their risk portfolio was small and manageable, soon found out that it wasn't.
There's a big difference between calculation and automation though. This article starts with an operation that's a bit of both, but the big underlying theme here is the automation of data digestion.
For all the pomp and pedigree of banking, it's mostly just moving data or digesting data. Because of how complex banks are, they're easily a decade behind the times and still HEAVILY reliant on manual processes in excel. Any real "streamlining" of processes is done through patchwork fixes with no systemic reworks being done.
This article is implying systemic reworks. Banks live and die on their NII, and would gladly gut their back and middle office personnel (most have done so by outsourcing to India and Hungary) with automation. Truth be told, rightfully so given the menial tasks of those jobs.
>> The program does the mind-numbing job of interpreting commercial-loan agreements
A computer program that interprets written language -- natural-language processing -- is reliant on highly-advanced, complex algorithms and AI to do the interpreting. This is much deeper than mindless "automation" and requires absolute trust in the accuracy of the underlying algo (or, hiring back some of those humans).
I would like to see what's behind the hood.
Since JPM is a bank I guess they used SAP HANA using the text analysis services... but I would like to know a bit more about the project.
According to this same Bloomberg article, JPM has 20k+ developers, I am fairly sure they develop a significant portion of the softwares they use internally, including their own cloud.
See the original article, 8th paragraph (under the sub heading "Redundant Software")
There has always been a financial-legal-accounting complex that has accounted for some large percentage of the economy. That used to be a large part of the labor-capital compact -- at least some of the proceedings of capital were going to well-priced labourers who could build the foundations of the upper-middle class or slightly above.
Capital and its holders are getting more and more naked in their attempt to save costs. Individually, this is great -- spurring automation forward certainly frees people up. In aggregate, however, if political/social factors aren't taken into account with automation, there will be a spate of people who think they will have been robbed of their rightful earnings and their purpose in life -- and that will have consequences, intended and unintended.
> That used to be a large part of the labor-capital compact
No, it was just a necessary cost of maximizing capital returns given the available tools. Elite labor often mistook itself for a natural class ally of the capitalist class because of this, but that was a false belief driven by transitional rather than fundamental conditions.
While this article seems to be a bit light on specifics, the automation of most legal practices is something that ML will have a great affect on in the near future (imo). Junior associates get paid insane salaries to do very meaningless grunt-work at biglaw firms. Those who aren't on track to become partners, are essentially burning the firms' resources, thus there exists an big incentive to automate the work.
Junior associates are not burning the firm's resources. They are the firm's resources. They generate billable hours, which is how the partners make money. Which is why most of the legal profession is much slower to replace them than you might think.
This is true too. The billable hour billing practice incentivizes inefficiency. In some respects that makes the industry ripe for disruption, but it's staying power may suggest the opposite, that is, that the business isn't about being efficient. Sometimes as logical people we assume everyone else is logical too. But my experience is that law firms are stuffed with worker bee types who don't see the world that way. They are not the innovative logical types who would develop something to increase efficiency
I think what JPMorgan fails to realise is that their entire business will be subject to automation. In a way they are signing their own death warrant. Obviously they will reap the benefits of AI while it's in it's infancy, but it won't take a quantum leap to realise that most of their business can be replaced by software eventually.
> Not all of those bets, which include several projects based on a distributed ledger, like blockchain, will pay off, which JPMorgan says is OK. One example executives are fond of mentioning: The firm built an electronic platform to help trade credit-default swaps that sits unused.
so it looks like JPM might have written off "blockchain"?
Some of my friends who are now lawyers were really bad at math in high-school, so I'm not sure that equating law to computer programs is how it works in practice.
It's not the math, it's logic. I see this parallel too -- legalese is incredibly dense because it is very specific, and its unreadability stems from (our) natural languages' disinclination towards specificity
To be fair, computer programming itself doesn't really require a lot of math knowledge. Conceptual understanding of the programs does. But Logic is something that is basically required in both fields, so it's natural to see some similarities.
The edge cases are also exactly when lawyers get involved. If the law is clear, there's not much work for a lawyer to do (you plead guilty, you settle the case, you don't sue in the first place, etc.). Dealing with ambiguity is the main skill of a lawyer.
Automation: less jobs, Government (at least in Europe/Cdn) say more people we need to import more people to fill the jobs... but millions of jobs will disappear over the next 10 yrs.
There will be a point where "software" will automate all of this stuff. 360,000 hours is a ton of full time employees. I wonder how they feel, and if they are typing up resumes as we speak. Looks like machine learning and similar skills are going to be more sought after than the normal finance or law skills!
With that in mind, a key quote for me in the article was this:
>To help spur internal disruption, the company keeps tabs on 2,000 technology ventures, using about 100 in pilot programs that will eventually join the firm’s growing ecosystem of partners.
I think there will be a role emerging within Enterprises, (sort of like an "insurance broker" who finds the best coverage options among a sea of choices), that will be responsible for surveying new technologies that could potentially help the firm, and running pilot programs to see how they potentially help make the org. more efficient or more competitive.
One could argue that this is really the CTOs role, and it is, but often CTOs are focused on building their own technology, or maintaining existing legacy systems, and not so much on finding and implementing new technology that has not yet hit the main stream. So, at least the focus needs to shift.
*edit: typos