4 hours ago, Retreadfed said:

Just how much do we want to rely on AI? Can you imagine the outcome of a bid protest challenging a source selection decision when the SSA says "I used AI to make the decision and just signed off on what it generated"?

People would remain the starting and end points for all such decisions. OpenAI might be helpful in cross-checking instructions to offerors and evaluation criteria against a company's response to an RFP. How about the simple act of using control-F as a function to search a .pdf or text-based document? What if I wanted to find all occurences of the term "best value" in the FAR. The "find" function would allow me to do this much more quickly than thumbing through the entire compendium. Still, I would need to read through the regulation and identify the specific paragraphs before and after each mention of the term to gain a more comprehensive understanding. Internet search engines are another example. I was raised on library card catalogs, the Dewey Decimal System, and microfilm—tools that were invaluable in shaping me into the researcher I am today. However, I wouldn't argue against the convenience of simply 'Googling' certain topics for market research or preparing for an upcoming presentation.

All this has me thinking about a couple books I recently read, from which I will quote:

"Teenagers today have to get greasy by working on old cars whose engines are simple enough for an amateur tinkerer to understand. Even professional mechanics have been heard to complain that they don't fix cars anymore; they just replace the modules that their computer tells them to replace" (The Knowledge Illusion: Why We never Think Alone; Steven Sloman and Philip Fernbach; p.27)

"Modern airplanes are so complicated that no one person completely understands them. Rather, different people understand different aspects of them. Some are experts on flight dynamics, others on navigation systems; several are required to understand jet engines; and some understand the ergonomics of seating well enough that companies are able to pack people into economy class with the same efficiency that Pringles are packed into a can." (Id., p.28)

"...computers do a much better job than judges at making bail decisions. The computer can't see the defendant. Judges can, and it seems logical that that extra bit of information ought to make them better decision-makers. Solomon, the New York State judge, could search the face of the person standing in front of him for evidence of mental illness—a glassy-eyed look, a troubled affect, aversion of the eyes. The defendant stands no farther than ten feet in front of him and Solomon has the chance to get a sense of the person he is evaluating. But all that extra information isn't actually useful. Surprised people don't necessarily look surprised. People who have emotional problems don't always look like they have emotional problems." (Talking to Strangers: What We Should Know About The People We Don't Know; Malcolm Gladwell; p.163-4)

4 hours ago, Don Mansfield said:

Does an SSA have to be human? If AI produced a rational source selection decision that would otherwise withstand scrutiny at the GAO or COFC, what's the problem?

That's a good question, @Don Mansfield.

Anyone who has used AI and attempted to incorporate it knows exactly how far away the technology is. Sure, at some point in the distant future there may be room for augmentation, but pretending like A.I. is a panacea or some wonder technology that people simply aren't using because they don't like change or don't want to become a horse is garbage. Every agency has Contracting Officers and Specialists who would love nothing more than to press the envelope because doing so would further their career

Yet the best you get are hilarious missives begging people to "just play around in GPT and see how it can help you! PLEASE FIND SOMETHING....ANYTHING...PLEASE!!"

Yet, crickets.

Even ACQbot churns out garbage. Are we going to be eliminating synopsis requirements because AI generated a list of potential vendors so why bother?

The reality is the technology can't even generate veritable requirements documentation (see:create,) let alone evaluate source selection information. The best it can do right now is write five bullets. Maybe it can write CPARS garbage that typically was copy/pasted in the past.

The days when the robo-GAO IS auditing SkyNet isn't likely in the next 30 years.

We can quote million books or articles about change but when the technology is oversold the discussion is pointless.

For the love of God, bring on FAR 2.0. Please. My username is from when I first started as an 1102 and after reading through the FAR, and all the ridiculous Supplements left me wondering, "why?" Let's even cut statutes. Example, the Small Business Act. Why is the Government in today's environment in the business of propping up "small businesses?" Hell, even the definition of a small business is not really describing a small business. 1,400 employees is small? Ha! Cut all of FAR Part 19, to include the non-manufacturing rule.

I worry for a lot of my peers when hopefully the fat gets cut and we are left with a fillet. Let us innovate and cook like Alton Brown does in a kitchen. My counterparts all want checklists and a repetitive process. Even for simple things like task orders below the MPT where we could train a crow to process those requests.

You gain experience through time or through reading about someone else's time. People need to pick up books and read more.

13 hours ago, Self Employed said:

Anyone who has used AI and attempted to incorporate it knows exactly how far away the technology is. Sure, at some point in the distant future there may be room for augmentation, but pretending like A.I. is a panacea or some wonder technology that people simply aren't using because they don't like change or don't want to become a horse is garbage. Every agency has Contracting Officers and Specialists who would love nothing more than to press the envelope because doing so would further their career

Yet the best you get are hilarious missives begging people to "just play around in GPT and see how it can help you! PLEASE FIND SOMETHING....ANYTHING...PLEASE!!"

Yet, crickets.

Even ACQbot churns out garbage. Are we going to be eliminating synopsis requirements because AI generated a list of potential vendors so why bother?

The reality is the technology can't even generate veritable requirements documentation (see:create,) let alone evaluate source selection information. The best it can do right now is write five bullets. Maybe it can write CPARS garbage that typically was copy/pasted in the past.

The days when the robo-GAO IS auditing SkyNet isn't likely in the next 30 years.

We can quote million books or articles about change but when the technology is oversold the discussion is pointless.

I have sat through demos of several procurement-focused AI platforms in recent days (some have yet to be released to the market). I was absolutely stunned at how accurate the output was. Whether it was an SOW or a certain kind of D/F, the docs weren't garbage at all. This is absolutely going to change how we do business. It's here. It's arrived.

20 hours ago, Don Mansfield said:

Does an SSA have to be human? If AI produced a rational source selection decision that would otherwise withstand scrutiny at the GAO or COFC, what's the problem?

Yes, an SSA has to be human. The duties and responsibilities of an SSA as described in 15.303(b) are inherently Governmental per 17.503(c)(12).

35 minutes ago, Motorcity said:

I have sat through demos of several procurement-focused AI platforms in recent days (some have yet to be released to the market). I was absolutely stunned at how accurate the output was. Whether it was an SOW or a certain kind of D/F, the docs weren't garbage at all. This is absolutely going to change how we do business. It's here. It's arrived.

Can confirm. It's not ready to "take over" but it is absolutely a viable tool to help reduce rote tasks. Now, to train the workforce so that we have a strong cadre of high level critical thinkers. A lot has been lost in favor of volume and "fast."

1 hour ago, ricroy said:

Yes, an SSA has to be human. The duties and responsibilities of an SSA as described in 15.303(b) are inherently Governmental per 17.503(c)(12).

Neither of those say anything about an SSA being human.

Do you think all award decisions are currently made by humans?

AI is well-suited to decision-making and will get better. It may already be better at making decisions than humans.

Look at the history of human decision making. The Kennedy, Johnson, and Nixon administrations were staffed by some of the best foreign policy and military minds in the world, and yet they decided to make war in Vietnam, even though some very experienced military advisors told them it would be mistake. And they decided to stay there for years, even after they realized by 1967 that we could not prevail militarily with the means we were willing to employ.

Ask the dead about the quality of human decision-making.

AI for source selection? Piece of cake.

5 minutes ago, Vern Edwards said:

AI is well-suited to decision-making and will get better. It may already be better at making decisions than humans.

Look at the history of human decision making. The Kennedy, Johnson, and Nixon administrations were staffed by some of the best foreign policy and military minds in the world, and yet they decided to make war in Vietnam, even though some very experienced military advisors told them it would be mistake. And they decided to stay there for years, even after they realized by 1967 that we could not prevail militarily with the means we were willing to employ.

Ask the dead about the quality of human decision-making.

AI for source selection? Piece of cake.

Yes, that is the thing - AI can learn and improve by adding quality documents/templates/samples to the pool of information. I suppose it will be up to humans to determine what is and what isn't a quality document. We may get to a point where the system is able to take in garbage and spit out gold. If it is going the way I think it will, we will need far fewer 1102s in the direct years ahead.

2 hours ago, ricroy said:

Yes, an SSA has to be human. The duties and responsibilities of an SSA as described in 15.303(b) are inherently Governmental per 17.503(c)(12).

Is there a statute that speaks to "inherently governmental functions"? Isn't FAR 7.5 derived from OFPP Policy Letters?

Oh, the humanity! Why couldn't a government official with delegable procurement authority delegate that authority to a computer program?

I don't advocate for automating away the 1102 series, but the question is interesting.

Edited April 8Apr 8 by MBrown
Typo - wrong FAR subpart citation

Just now, MBrown said:

I don't advocate for automating away the 1102 series, but the question is interesting.

The automation of some key 1102 functions is inevitable.

You will have to be a master in order to survive and prosper.

47 minutes ago, Vern Edwards said:

AI is well-suited to decision-making and will get better. It may already be better at making decisions than humans.

Hmmm. AI was invented by humans but can make a better decision than a human? I will always question the decision of humans whose track record for the purposes of AI are what create an AI decision. From the eye of this beholder.

2 hours ago, Don Mansfield said:

Neither of those say anything about an SSA being human.

Do you think all award decisions are currently made by humans?

15.303(a) states, "The contracting officer is designated as the source selection authority, unless the agency head appoints another individual for a particular acquisition or group of acquisitions." An individual is defined as a "a single human being as distinct from a group, class, or family." So as currently described in the FAR, an SSA is a human. Perhaps the new and improved FAR 2.0 will redefine "Contracting Officer" or "individual" as other than human.

--

The Federal Activities Inventory Reform Act of 1998 defines an inherently Governmental Function as "a function that is so intimately related to the public interest as to require performance by Federal Government employees," (emphasis mine) so this is based upon statute rather than any policy or regulation.

I also do not think that computer software could yet be considered a "Federal Government employee."

--

The question of whether or not all award decisions are currently made by humans is unimportant as the law requires humans to perform this work.

Edited April 8Apr 8 by ricroy

1 hour ago, C Culham said:

Hmmm. AI was invented by humans but can make a better decision than a human? I will always question the decision of humans whose track record for the purposes of AI are what create an AI decision. From the eye of this beholder.

I think in the realm of procurement, yes AI will be able to make solid and effective decisions. These are complex systems and while the FAR is complex, it certainly isn't rocket science.

1 hour ago, ricroy said:

15.303(a) states, "The contracting officer is designated as the source selection authority, unless the agency head appoints another individual for a particular acquisition or group of acquisitions." An individual is defined as a "a single human being as distinct from a group, class, or family." So as currently described in the FAR, an SSA is a human. Perhaps the new and improved FAR 2.0 will redefine "Contracting Officer" or "individual" as other than human.

--

The Federal Activities Inventory Reform Act of 1998 defines an inherently Governmental Function as "a function that is so intimately related to the public interest as to require performance by Federal Government employees," (emphasis mine) so this is based upon statute rather than any policy or regulation.

I also do not think that computer software could yet be considered a "Federal Government employee."

--

The question of whether or not all award decisions are currently made by humans is unimportant as the law requires humans to perform this work.

Where did you get your definition of individual?

FAR 7.503(c)(12) doesn't include selecting sources.

AI reality check.

OMB, today "Accelerating Federal Use of AI through Innovation, Governance, and Public Trust" Chief AI Officers. Within 60 days of the issuance of this memorandum, the head of each agency must retain or designate a Chief AI Officer (CAIO). CAI Os will promote AI innovation, adoption, and governance, in coordination with appropriate agency officials. Agency heads may choose to designate an existing official, such as a Chief Information Officer, Chief Data Officer, Chief Technology Officer, or similar official...For CFO Act agencies, the CAIO must hold a position at the Senior Executive Service, Scientific and Professional, or Senior Leader level, or equivalent.

The administration seems to be working at cross purposes here. My agency recently fired its Chief Information Officer, Chief Data Officer, Chief Technology Officer, along with their offices and much more - maybe half of our 30 SESs. My agency has also been told to plan for a 40% reduction in IT contract expenditures, which I understand is typical for other agencies within my department. So AI isn't happening for us.

Didn't you hear? AI in its current form is a magical buzzword that solves all issues so you need less staff. You'll just click the right NAIICS/FSC/PSC and AI will handle all contracts flawlessly while you're in your flying car. Any changes to specifications or interaction with the lowly humans bidding on your efforts will be flawlessly adjudicated by NOPEGPT. NOPEGPT additions will also be added to handle customer interaction/education/changes, with retina-scans so you can record in real time the exact moment they lose faith in actually getting what they want. NOPEGPT will also be programmed to negotiate directly with LM/Boeing bots, and you will now book travel for the bots to meet in person so they can argue with each other on the same network to reduce latency -- instead of protests/claims we'll solve it all with battlebot matches. There won't ever be any issues as we roll out to staff who struggle with excel and everything that can go wrong and currently needs human eyes/ears to resolve will simply be consolidated to two or three curmudgeons made available in-between their hourly cans of Brawndo. Be on the lookout for J&ABot, where you can check in real time as useless revisions/bloat are added now by bots instead of people.

This is the first article that provides at least a few details including various increases and processes. I am curious to know about this user guide that they plan to release (plus all the threshold increases).

Washington Technology

Plan for sweeping FAR changes nears release

The General Services Administration, NASA and Office of Federal Procurement Policy are working to create a leaner acquisition framework that could attract more companies to the market.

On 4/8/2025 at 11:53 AM, Vern Edwards said:

The automation of some key 1102 functions is inevitable.

You will have to be a master in order to survive and prosper.

Any system whether in the federal government or outside of it that requires everyone to be a master is unrealistic and doomed to fail. The existence of masters presumes the existence of nonmasters otherwise the word would have little meaning.

Should we shoot for mastery? Yes. Should the FAR be rewritten such that only masters can be effective in using it? Human nature and the history of the world say the answer is No.

1 minute ago, TakeAWalkontheFARSide said:

Any system whether in the federal government or outside of it that requires everyone to be a master is unrealistic and doomed to fail.

@TakeAWalkontheFARSide

I agree.

Hopefully, no one expects the FAR to be rewritten so that only masters can be effective. (Masters are more than just effective.)

Most of the people on the DAR Council and the Civilian Agency Acquisition Council are not maters. I had a friend on the DAR Council, now dead, who told me that there was not a person on the DARC who could write a decent English sentence. But the FAR should be written so that journeymen can effective.

But I would like to think that all contracting officers with unlimited warrants are masters, though even that is expecting too much from the government.

I'm working on an article about mastery for publication later this year. I hope it will be posted at Wifcon. It's an interesting topic.

Is the FAR being updated specifically so that much of the procurement process can be automated? I do wonder if automation was even taken into consideration during the streamlining of the FAR. I guess what I am asking is will the updated version of the FAR be automation friendly?

7 minutes ago, Motorcity said:

I guess what I am asking is will the updated version of the FAR be automation friendly?

Don't know.

2 hours ago, Motorcity said:

first article that provides at least a few details

From article "Work is also underway on Part 34..."

I am looking forward to this. For such a short Part, it needs a lot of work.

Part 34 describes "acquisition policies and procedures for use in acquiring major systems consistent with OMB Circular No. A-109." The elusive OMB Circular A-109. I challenge you to find this document online and provide a link to it. Go head, search for it. Spoiler: It is very hard to find. Nobody has read this circular in decades. Wasn't it superseded by A-11 in, like, the 1990s (A-11 has a 75-page guide titled "PLANNING, BUDGETING, AND ACQUISITION OF CAPITAL ASSETS", which surely is the correct citation)? Why wasn't the FAR updated long ago?

Why is EVM required, ever? Does anyone actually do EVM? Maybe they do and find it very useful. I honestly don't know, but doubt it.

Why is there a section about competition in FAR 34? Isn't this covered elsewhere, like, for example FAR Part 6?

Why does my department tell me I must discuss the non-applicability of pre-award IBR for every contract action over the SAT, most of which are a few orders of magnitude removed from being 'major'? Why especially given my department has de facto outlawed pre-award IBRs?

4 minutes ago, General.Zhukov said:

Part 34 describes "acquisition policies and procedures for use in acquiring major systems consistent with OMB Circular No. A-109." The elusive OMB Circular A-109. I challenge you to find this document online and provide a link to it. Go head, search for it. Spoiler: It is very hard to find.

It's not hard to find. You just have to know how and where to search. Google Books:

Hearings on Office of Management and Budget Circular A-109, Major system Acquisition Policy, before the Research and Development Subcommittee of the Committee on Armed Services, House of Representatives, Ninety-Fifth Congress, First Session, November 2, December 1, 1977, Second Session, April 6, 10, and 14 and September 18, 1978.

https://books.google.com/books?id=6Gvef5JFImUC&pg=PA545&dq=hearings+on+OMB+Circular+A-109+major+systems+acquisition+policy&hl=en&newbks=1&newbks_redir=0&sa=X&ved=2ahUKEwiRnafZgs6MAxVpATQIHct2IaoQ6AF6BAgIEAM#v=onepage&q=hearings%20on%20OMB%20Circular%20A-109%20major%20systems%20acquisition%20policy&f=false

The full text of the 15-page circular begins on page 4.

I thought everybody knew that. 😂

39 minutes ago, General.Zhukov said:

Nobody has read this circular in decades.

That's not true. I had a discussion about it with a DAU professor not too long ago. It has been mentioned in the Federal Register twice since 2021.

To the best of my knowledge OMB A-109 has never been rescinded. In 2000, OMB proposed the "recission" of A-109, see 65 FR 51045-01, August 20, 2000, but apparently did not follow through.

In 2006 the FAR councils confirmed the continuing applicability of OMB A-109:

Comment: Some commenters suggest removing references to OMB Circular A-109 in FAR Part 34.000. No reason was provided.

Response: The Councils believe that OMB Circular A-109 continues to apply. OMB Circular A-11, Part 7 supplements OMB Circular A-109, which has not been rescinded by OMB, and is still available. Therefore, a reference to OMB Circular A-11 has been added to FAR 34.000.

71 Fed. Reg. 38238-02, July 5, 2006.