Is Artificial Intelligence a Solution to Contracting Problems?

AI can facilitate and speed the acquisition process whenever a decision of any kind must be made that can be made on the basis of an algorithm. Every facet of acquisition can be structured in terms of a decision making process--contract type selection, CLIN structuring, clause selection, proposal evaluation, source selection, determinations of cost allowability, determinations of price reasonableness, claim decisions, etc.

It's clear that the people we now call COs will be mainly data processing techs in the not-too-distant future. Certainly within a generation, if that's what management wants.

This was discussed in the AT&L Magazine back in 2015:

http://dau.dodlive.mil/2015/01/12/a-contract-that-manages-itself-the-time-has-arrived/

Matthew - I have for years thought that software can do the bulk of what a CO can, and it can do it faster, cheaper, and more accurately. Of course, there would still be points along the course of an acquisition that require human input, and more complex acquisitions might require greater human involvement, but I believe that automation is the future of the contracting profession.

Having read many of your comments here at Wifcon, and knowing how knowledgeable you are, I'm actually really surprised that you don't see merit in this approach. I hate to come across as overly-negative, but truly, quality contracting personnel are an exception (in the DC region, anyway). You need only to pull a random contract file or try to hold a FAR-based conversation with a random CO, and you will likely see that this is the case. And no amount of training or "materials" will change this, both of which require a willingness on the part of the participant to work. (If you want to improve the workforce with people, raising salaries and threatening jobs are better starting points in my opinion.)

(I want to emphasize that I have met numerous 1102s in my career who I consider to be strong. But they are significantly outnumbered by the weak ones.)

First there will have to be a huge investment in the IT budget.  We still don't even have an electronic offer system in my service - to go to full AI it will take that and a LOT more.  The biggest hurdle would be on best value tradeoff decisions, but there's ways (some for better, some for worse) you could automate that as well.

Best value tradeoff decisions would be easy to automate. Much easier than chess moves, which, at championship level of play, are much more demanding.

PepeTheFrog, for one, welcomes our new AI contracting overlords. Robocop would make better decisions than the average GS-13 contract specialist. 

Automation, AI, algorithms, and machine learning will bring us closer and closer to the warm and fuzzy goal of corporate feudalism.

Our betters (some call them "leaders" or "elites") want us to import millions of low-skilled peasants at the same time as we hurtle toward automation and new frontiers of technology.

What could go wrong?

Who needs a coherent and unified culture, a shared sense of civic duty, or a national character when you can efficiently farm humans with a Universal Basic Income and an entertainment and spying device in every pocket and bedroom? 

Only a Luddite or xenophobe would object. Progress!

PepeTheFrog said:

PepeTheFrog, for one, welcomes our new AI contracting overlords. Robocop would make better decisions than the average GS-13 contract specialist. 

Automation, AI, algorithms, and machine learning will bring us closer and closer to the warm and fuzzy goal of corporate feudalism.

Our betters (some call them "leaders" or "elites") want us to import millions of low-skilled peasants at the same time as we hurtle toward automation and new frontiers of technology.

What could go wrong?

Who needs a coherent and unified culture, a shared sense of civic duty, or a national character when you can efficiently farm humans with a Universal Basic Income and an entertainment and spying device in every pocket and bedroom? 

Only a Luddite or xenophobe would object. Progress!

Where might one subscribe to your newsletter?

@PepeTheFrog "alexa, write me a task order!"

Here’s some more food for thought on AI from this news this week:

https://www.washingtonpost.com/business/economy/ais-ability-to-read-hailed-as-historical-milestone-but-computers-arent-quite-there/2018/01/16/04638f2e-faf6-11e7-a46b-a3614530bd87_story.html?utm_term=.f12ad2e2acde

Quote

“Technically it’s an accomplishment, but it’s not like we have to begin worshiping our robot overlords,” said Ernest Davis, a New York University professor of computer science and longtime AI researcher.

Maybe the robots will make their humans “pets” someday.

I was reminded of this discussion while reading Thinking, Fast and Slow recently (an excellent book on human psychology and behavior, for those who haven't read it). In it, there is a chapter on Intuition vs. Formulas. The takeaway is that formulas and algorithms are usually far better at judging situations and predicting outcomes than human intuition for two primary reasons:

1. people tend to over-complicate matters; and
2. people can have vastly different conclusions about the same set of data because of innumerable variables having nothing to do with the subject at hand, including minute environmental differences that influence perception.

The book goes on to state that intuition can add value in some circumstances, particularly if the data on which the judgment is based is objective and predictable. 

I also think another chapter in the book, Illusion of Validity, informs this discussion. Humans tend to have a subjective belief in their judgments and predictions, even when there is objective evidence of low correlation between judgment and outcome (e.g., professional stock traders). When we are confident in our judgment, it is typically because our mind has constructed a sufficiently coherent narrative, which may or may not conform to reality.

Considering this, it seems utterly reckless to me to have tens of thousands of COs, whose qualifications are varied and typically not strenuous, making decisions on behalf of our Government based on personal discretion and interpretation of a regulation renowned for its flexibility. Bring on Skynet, I say. Worth the risk.

I've become a believer in AI, too, after reading Supercrunchers by Ian Ayres. I would like to see the use of a nonhuman source selection authority in my lifetime.

Don Mansfield said:

I've become a believer in AI, too, after reading Supercrunchers by Ian Ayres. I would like to see the use of a nonhuman source selection authority in my lifetime.

Did the book just talk about AI's ability to compute/make decisions or did it also include any worthwhile discussions of AI's ability to learn?  I'm more familiar with the former than the latter so if you give it a solid recommendation I'm going to put it on my reading list.

There were worthwhile discussions of the ability to learn. Fascinating discussion of the use of neural networks to solve different problems. However, the book is 10 years old so I suspect the technology is more advanced today. I still recommend it.

Don Mansfield said:

I would like to see the use of a nonhuman source selection authority in my lifetime.

Don,

You may want to reach out to the folks at DLA. DLA is reportedly awarding 3K orders per day through automation that includes an automatic evaluation system.

Quote

DLAD 13.003-90 Policy.

(e) It is the policy of DLA that each contracting office shall use the automated solicitation, evaluation, and award processes to the fullest extent possible. Any requirements that are not satisfied as delivery orders against long-term contracting arrangements are candidates for automated procurement.

Thanks, Jamaal.

Don Mansfield said:

I've become a believer in AI, too, after reading Supercrunchers by Ian Ayres. I would like to see the use of a nonhuman source selection authority in my lifetime.

Don, you're reading too much of that kind of crap. Life is too short. There are wonderful new translations of The Three Musketeers and The Odyssey. Would you like me to send you a copy of one of them? Or maybe you just want a little plastic wind-up Robby the Robot toy for your desk at DAU. Or maybe a DVD of "Colossus: The Forbin Project."

Jamaal Valentine said:

Don,

You may want to reach out to the folks at DLA. DLA is reportedly awarding 3K orders per day through automation that includes an automatic evaluation system.

That is really cool. Wonder why there isn't more buzz about this (positive or negative). Any word yet on results or trends?

Vern Edwards said:

Don, you're reading too much of that kind of crap. Life is too short. There are wonderful new translations of The Three Musketeers and The Odyssey. Would you like me to send you a copy of one of them? Or maybe you just want a little plastic wind-up Robby the Robot toy for your desk at DAU. Or maybe a DVD of "Colossus: The Forbin Project."

I'm going through a mid-life crisis, ok? Some guys go out and buy sports cars to feel young, I read books about math. I used to want to be an actuary. That is my road not taken.

Math! That's a high class midlife crisis. I wanted to be in the Foreign Legion. (It's a long family history story.) Let myself be talked out of it. Ended up in contracting.

What a fail. 

On ‎4‎/‎20‎/‎2018 at 11:13 AM, Jamaal Valentine said:

Don,

You may want to reach out to the folks at DLA. DLA is reportedly awarding 3K orders per day through automation that includes an automatic evaluation system.

DLAD 13.003-90 Policy.

(e) It is the policy of DLA that each contracting office shall use the automated solicitation, evaluation, and award processes to the fullest extent possible. Any requirements that are not satisfied as delivery orders against long-term contracting arrangements are candidates for automated procurement.

Since DLA is on the cutting edge, I wonder if they have a plan...if things go badly, just how does an AI get sentenced to Leavenworrth?

On 4/19/2018 at 10:13 PM, Jamaal Valentine said:

Don,

You may want to reach out to the folks at DLA. DLA is reportedly awarding 3K orders per day through automation that includes an automatic evaluation system.

As one who helped DLA set up the contracting system component of the ERP which formed the  basis of this automation, I can tell you that DLA is fairly unique in its high volume/low dollar/low complexity business model. In fact, much like the Obamacare website, things didn't go well upon go-live because PD2 choked on the volume of orders and line items.

DLA also has a system of Prime Vendors who hold long term ordering vehicles that facilitate this automated model, and the agency is organized around firmly segregated and predictable lines of business (e.g., land, maritime, air).

It's also worth noting that DLA built a manual-assist process from the get-go, which kicked out a significant # of nominally simple procurements which were too complex for automated processing (which were identified by a separate PIID schema).  So even the automated process for a commodity-focused procurement system was built with the expectation that everything couldn't be automated

One can rebuff artificial intelligence all he wants.  The fact is that technology continues to evolve.  The United States was primarily agricultural through the Civil War, with approximately three-quarters of the population living in rural areas.  The ensuing transition toward industrialization and urban migration inevitability led to specialization within the larger workforce.  Dare we forget that American manufacturing turned the tide in World War, accelerating the defeat of the Axis powers and lending to the rebuilding of swaths of Europe and the Far East.  We maintained our status as the world’s industrial powerhouse until about the late 70s or early 80s when manufacturing jobs began to move overseas.  Today, we identify as a service economy.  The most purchased commercial items in dollars are exempted from domestic manufacturing requirements by the Buy American Act.  Our most profitable supplies are those of the intellectual property type.  The creation of software applications and like products is wholly reliant on the most exclusive forms of specialized knowledge and creative risk-taking.  Even our service industry has been turned on its head.  Today, the world’s largest provider of lodging owns not a single hotel.  The largest passenger transportation service owns no vehicles.  The majority of freight companies are brokers, subcontracting to trucking companies of varying sizes.  The cost of freight in many regions of the county has nearly doubled, even though pay to drivers is at a historic high.  With Boomers retiring, the need for professional drivers is greater than ever.  But Millennials and Generation Z are not seeing the allure of a job relegating them to 11-hours a day of confinement.  They know too that they would be lucky to see the maturation of another generation before the entire industry transitions to driverless vehicles.  The same should be a concern for the 160,000 Uber drivers in the US.

You can like it or hate it, AI is indifferent.  AI may be in its budding infancy, but in time, it will surely grow to colossal proportions.  We use predictive modeling and algorithmic-based applications on a limited scale in my office and I for one could not be happier.  We are awarding a $147M supply contract containing over 4700 line items.  Each represents a truckload delivery to an individual recipient.  The final destinations span all 50 states, certain territories and Indian reservations.  Offerors propose prices per state (or territory) for each item on the schedule of supplies.  Offerors may indicate constraints in their proposals, i.e., limitations to particular locations and maximums for particular items.  Apply these “constraints” to 4700 some odd items and it becomes the kind of price evaluation that my limited and impatient mind would rather not perform.  The algorithmic software we use allows us to enter rules up front, including set-asides, then sorts the data per our instruction.  The result is the creation of a scenario, complete with bar and pie graphs and several analytical representations of the data.  Why have a protest hinge on the precision of your own calculus when you can rely on your handy TI-84 Plus?

I agree that many specialists are consigned to repetitive tasks and vapid clerical work.  The larger injustice lies in circumstances in which a specialist lacks capability or is unwilling to do more, but just the same benefits from the gainful rewards in pay and other incentives.  Perhaps even more unfortunate is the skillful and ambitious CS who struggles to remain challenged and stimulated.  Every organizational shift ought to contain the seeds of an equivalent or greater benefit.  I would like to say that the smart ones will adapt, but intelligence is only part of it.  There too has to be a willingness and fit determination.

AI can’t vote or run for office...it’s part way to felon status right out of the box...

Here's a story about a kind of low-grade artificial intelligence at work:

http://www.newsweek.com/publix-summa-cum-laude-sheet-cake-938451

I think government contracting is a trivial concern when it comes to AI. But see the current issue (June 2018) of The Atlantic. It contains an interesting article about artificial intelligence by Henry Kissinger, "How the Enlightenment Ends," which is a little unnerving and might give one pause.

In the context of this thread, consider this quote and its implications for source selection decision making:

Quote

AI may reach intended goals, but be unable to explain the rationale for its conclusions. In certain fields—pattern recognition, big-data analysis, gaming—AI’s capacities already may exceed those of humans. If its computational power continues to compound rapidly, AI may soon be able to optimize situations in ways that are at least marginally different, and probably significantly different, from how humans would optimize them. But at that point, will AI be able to explain, in a way that humans can understand, why its actions are optimal? Or will AI’s decision making surpass the explanatory powers of human language and reason? Through all human history, civilizations have created ways to explain the world around them—in the Middle Ages, religion; in the Enlightenment, reason; in the 19th century, history; in the 20th century, ideology. The most difficult yet important question about the world into which we are headed is this: What will become of human consciousness if its own explanatory power is surpassed by AI, and societies are no longer able to interpret the world they inhabit in terms that are meaningful to them?

How is consciousness to be defined in a world of machines that reduce human experience to mathematical data, interpreted by their own memories? Who is responsible for the actions of AI? How should liability be determined for their mistakes? Can a legal system designed by humans keep pace with activities produced by an AI capable of outthinking and potentially outmaneuvering them?

Mark Zuckerberg made hoped for further developments in AI the centerpiece of his deflections while appearing before Congress last month. When pressed on FB’s own enforcement of community standards...his invocation of AI as a prospective cure seemed to me at best a transparent attempt at political alchemy and at worst an inversion of rational standards of trust. I hope that such superficial blandishments are not a harbinger of  things to come. Issues over commonplace AI technology do not strike me as a trivial...these issues call for great care in what we cede knowingly or not, to AI. We may not get a redo in our lifetime. The feedback loop is necessarily the most important part of any  iterative system...Just what will happen should AI become “self-interested” and already have a role in many feedback processes?   We hold to a chain abounding in weak links...will we catch them all before any one of them breaks...and who is doing the watching?