Jump to content

Magic Wand


BrettK

Recommended Posts

  • Replies 121
  • Created
  • Last Reply

Top Posters In This Topic

I tested the Contract Advice Bot. The answers were, shall we say, interesting.

I asked it several questions. It returned answers very quickly. The answers were primitive. 

For instance, I asked: "What is the proper measure of an equitable price adjustment?"

It answered:

Quote

 

As a United States Contracting Officer, the proper measure of an equitable price adjustment is based on the changes in the cost of labor, materials or performance requirements that were provided for in the initial contract. An equitable price adjustment ensures that the contractor is fairly compensated for changes that are outside of their control.

The Federal Acquisition Regulation (FAR) reference that relates to this is FAR 52.243-1, which outlines the requirements for changes and provides guidance on how to determine an equitable adjustment. This FAR clause requires the contractor to submit a proposal for any adjustment to the contract price, and it provides guidance on the factors that should be considered in determining the equitable adjustment.The Contracting Officer must then review the proposal and negotiate a fair and reasonable adjustment.

 

Educated Wifcon readers will consider that answer to be entirely unsatisfactory.

Of course, the Contract Advice Bot is just a start. I think it would be possible to develop a worthwhile advice Bot. But it would take a lot of time and cost a lot of money. Maintaining it would cost even more money.

And our government is having troubles with SAM registration.

 

Link to comment
Share on other sites

8 hours ago, Vern Edwards said:

I think it would be possible to develop a worthwhile advice Bot.

I think even a really great advice Bot will suffer from the same problem as DAU's Ask a Professor: A simple authoritative statement, even if all the references/logic/justification are included, doesn't contribute to understanding an issue in the same what that the back-and-forth discussion in this forum does (for participants or even just readers).

I read about a study recently that showed users gained a better understanding of an issue via traditional Google search, where they have to review and determine the validity of the search results themselves, than via an AI search which simply provided them "the answer" to their question without requiring any thinking on their part.

Link to comment
Share on other sites

8 hours ago, Vern Edwards said:

And our government is having troubles with SAM registration.

As already demonstrated by my few posts I am leery.   Looking into the matter of AI, across the spectrum, not just acquisition, one common thread keeps cropping up security.   The issue is mentioned in the Strategic Plan I posted earlier a couple of times.  I got engrossed in a read about it with regard to what I will call voice robbing, or in other words it is not really Carl speaking but a Bot but it sure sounds like Carl.   Can or does the same happen to Bot writing? 

Getting SAM right is a good example along with the other automated systems that are discussed in Forum that create solicitations and frustration.  I used to fix problems with a stubby pencil and eraser and as noted person to person discussion.  Now in some cases it takes almost extraordinary computer manipulation to make say a payment fix.  Then I circle back to security.  Heck even the most important thing to me today, OPM and my retirement, where the OPM data breach that started in 2013 took until last year to settle (for $63M by the way).  

I appreciate the enthusiasm for Bot solutions for acquisition.  I will be interested not only to see what the initial investment in time and money will be to make something almost completely workable to almost everyone's satisfaction  Then there is the evolution to what happens when a data breach occurs and is blamed for a contract mishap of any kind.   What then?

 

Link to comment
Share on other sites

31 minutes ago, C Culham said:

I used to fix problems with a stubby pencil and eraser and as noted person to person discussion.  Now in some cases it takes almost extraordinary computer manipulation to make say a payment fix. 

Carl, to quote Tennyson, "You and I are old."

My wife complains about the profanity coming out of my office now and then that is directed at one of my several "devices." Recently, my iPhone stopped ringing when someone called. I don't know why. I missed several calls. I tried everything to figure out how to fix the problem. Went online and all that. Did everything I was told to do. No luck. My nephew was over on Sunday and I told him about the problem. He asked for the phone and handed it back to me in about 20 seconds. Then he called me and it rang. It's been ringing ever since. I have no idea what he did. Then he kissed me on the top of my head.

The future is for the young.

Link to comment
Share on other sites

9 hours ago, Vern Edwards said:

Of course, the Contract Advice Bot is just a start. I think it would be possible to develop a worthwhile advice Bot. But it would take a lot of time and cost a lot of money. Maintaining it would cost even more money.

 

1 hour ago, Witty_Username said:

I read about a study recently that showed users gained a better understanding of an issue via traditional Google search, where they have to review and determine the validity of the search results themselves, than via an AI search which simply provided them "the answer" to their question without requiring any thinking on their part.

The future of AI likely will have the bot continuously updating and maintaining itself.  I think the vision is have it programmed initially by “experts” in the field to critically “think” and “analyze” all available data for itself and come up with the best answers.  Improvements will evolve over time and use.

Link to comment
Share on other sites

2 hours ago, Vern Edwards said:

Carl, to quote Tennyson, "You and I are old."

My wife complains about the profanity coming out of my office now and then that is directed at one of my several "devices." Recently, my iPhone stopped ringing when someone called. I don't know why. I missed several calls. I tried everything to figure out how to fix the problem. Went online and all that. Did everything I was told to do. No luck. My nephew was over on Sunday and I told him about the problem. He asked for the phone and handed it back to me in about 20 seconds. Then he called me and it rang. It's been ringing ever since. I have no idea what he did. Then he kissed me on the top of my head.

The future is for the young.

Yes, yes it is!  I do wonder if the future is as easy as they think it is going to be?

Link to comment
Share on other sites

8 hours ago, C Culham said:

OPM data breach that started in 2013

8 hours ago, C Culham said:

voice robbing, or in other words it is not really Carl speaking but a Bot but it sure sounds like Carl

This is going to really screw with people in the formative years of A.I.  By example I just watched a video on YouTube of a speech Thomas Sowell made in 1999.  I didn’t verify anything about the date or transcripts - just took the video title’s word for it.  Wait till A.I. starts posting these videos too.  Who’s going to watch every last word for a bit of ideological tweaking by an A.I. that is just “voice robbing” the speaker?  Nefarious actors could do a lot of social engineering in this way.  By the time a meticulous viewer realizes it’s fake, the masses will have already received an illegitimate message.  And in today’s day and age the masses won’t research beyond an expert’s (supposed) opinion.

Now, what if a nefarious state actor’s A.I., knowing all of a federal employee’s SF 86 contents (via the OPM data breach), started placing dead-air calls to that fed’s phone number on the form, and gained not only their past 7 years of personal information but also the ability to mimic the fed’s voice?

Link to comment
Share on other sites

9 hours ago, Witty_Username said:

I think even a really great advice Bot will suffer from the same problem as DAU's Ask a Professor: A simple authoritative statement, even if all the references/logic/justification are included, doesn't contribute to understanding an issue in the same what that the back-and-forth discussion in this forum does (for participants or even just readers).

I read about a study recently that showed users gained a better understanding of an issue via traditional Google search, where they have to review and determine the validity of the search results themselves, than via an AI search which simply provided them "the answer" to their question without requiring any thinking on their part.

Taking my tin foil hat off for this post, I return to something I learned from a book about curiosity that @Vern Edwards recommended in this forum, which explained that, thanks to Google, most people think they can treat all their problems as puzzles, with a definite answer.  These people fail to realize many problems they query Google to answer are actually mysteries, which require judgments be built upon judgments to arrive at a defensible, but uncertain, conclusion.  In other words, they require the application of research.

I don’t know about you, but I plan to strategically direct my career towards positions requiring research application to impress the boss, e.g., complex services or systems contracting.  The puzzles of our career field will just be solved by A.I. someday anyway.

Link to comment
Share on other sites

15 hours ago, WifWaf said:

By the time a meticulous viewer realizes it’s fake, the masses will have already received an illegitimate message.

Now I know that most people have the knowledge, skills and abilities to sort through "fake".   I do find it interesting in the last two days that the following emails have arrived in my email Inbox.

The latest from "donotreply@sam.gov" notifying me that (this was the subject of the email) "Someone updated the entity registration for _ _ _ _ _ _(name).   Several links in the email to the supposed "Entity Administrator" who made the changes and fake (as determined by me) links to SAM.gov to contact to dispute the changes.

And

An email from (supposedly) Miguel Cordona, Sec of Ed himself, inviting me to quote on an attached RFQ.  

Neither of these emails came through as spam but direct to my email address.

I just scratch my head, while also agreeing that AI will have its place, at what just my and others emails, none the less the rest of the communication world, will look like in future.  You see there are people out there that will not sort through "fake".   By the FTC's account 5.7 million folks were subject to fraud to the sum of $5.8 Billion in 2021.    https://www.ftc.gov/news-events/news/press-releases/2022/02/new-data-shows-ftc-received-28-million-fraud-reports-consumers-2021-0 

Not a reason to block AI but a dang good reason to wade in slowly.

And, yes I am old no doubt about it.

Link to comment
Share on other sites

17 hours ago, WifWaf said:

This is going to really screw with people in the formative years of A.I.  By example I just watched a video on YouTube of a speech Thomas Sowell made in 1999.  I didn’t verify anything about the date or transcripts - just took the video title’s word for it.  Wait till A.I. starts posting these videos too.  Who’s going to watch every last word for a bit of ideological tweaking by an A.I. that is just “voice robbing” the speaker?  Nefarious actors could do a lot of social engineering in this way.  By the time a meticulous viewer realizes it’s fake, the masses will have already received an illegitimate message.  And in today’s day and age the masses won’t research beyond an expert’s (supposed) opinion.

Now, what if a nefarious state actor’s A.I., knowing all of a federal employee’s SF 86 contents (via the OPM data breach), started placing dead-air calls to that fed’s phone number on the form, and gained not only their past 7 years of personal information but also the ability to mimic the fed’s voice?

Funny you should mention that.

There are unclass calls for deepfakes created by A.I. that already exist but not for non-commercial purposes, and DoD is interested. (4.3.1.4.)

 

 

AI is a critical future technology that is necessary as unmanned platforms become more prevalent.

Just as with DARPA, these military applications will trickle down into civil applications, to include augmenting our work capabilities as an 1102. Some will trickle. AI will pressure wash.

While it would be nice to be able to "take it slow," the ongoing great power competition unfolding will not wait. AI is a technology that our near peer competitors are in some respects "ahead," of our capabilities. The U.S. controls announced last year have expanded beyond the U.S. alone. The computing power required to sustain AI is playing out in a digital arms race in real time, globally.

GAO article

AI%20graphic.jpg.webp?itok=j3829SIq

 

Just with other innovative technologies of their time, some might cling to other ways/means -- but those that do will see their capabilities suffer. I don't think it's unreasonable to view AI's implementation as very soon, and likely sudden. 

Link to comment
Share on other sites

1 hour ago, Self Employed said:

Video I was describing watching was just a voiceover of Mr. Sowell talking while the uploader doodled to illustrate the speech.  It's the format that many education-minded YouTube accounts ascribe to.  Deepfakes need not be developed for these videos to be manipulated - only voice robbing A.I. is necessary.  That technology is already used widely.

Viewer beware of your sources.

Link to comment
Share on other sites

While the GAO report and SOCOM BAA don’t specifically address our contracting field, it doesn’t take much thinking to see how the would is evolving and that contracting will be radically different in the future.  I mentioned earlier trying to speculate what the specific impacts on our field is likely fruitless because things are moving so rapidly.  Some initiatives will be successful while others won’t but I think it’s safe to say what we now do as current contracting practices won’t work in a few years.

There are two routes 1102s can take - resist until they drag you out of your office, or adapt and be part of the change movement.  There are likely rewards for those that do the later.

 

Link to comment
Share on other sites

2 hours ago, formerfed said:

 

There are two routes 1102s can take - resist until they drag you out of your office, or adapt and be part of the change movement.  There are likely rewards for those that do the later.

 

Bingo.

Do the best you can within the confines of the system.

I look forward to real bullets being written about innovation in our field utilizing and incorporating these emerging technologies.

 

 

Link to comment
Share on other sites

On 3/8/2023 at 12:21 PM, Self Employed said:

Just with other innovative technologies of their time, some might cling to other ways/means -- but those that do will see their capabilities suffer. I don't think it's unreasonable to view AI's implementation as very soon, and likely sudden. 

We should not conflate advances in AI broadly (or in other domains) with advances in AI specifically related to government contracting - AI advancements requires effort and investment. Effort and investments usually follow the market - Industry is aware of the opportunity costs of their efforts and capital...what's the business case for pursuing AI in the government contracting field when there are likely considerably more lucrative opportunities available?

I think the market for a government contracting AI solution is quite small (we can barely scrounge together the pocket change out of the DoD budget to pay for CON-IT). And even if we do find the money, I have used enough government specific software to know that, unfortunately, we often don't end up with great solutions (and we certainly do not sustain them well).

All that said, I would love to be proven wrong by demonstrated capabilities, but most of what I've seen relative to AI applications to government contracting is either hope or hype.

Link to comment
Share on other sites

3 hours ago, Matthew Fleharty said:

We should not conflate advances in AI broadly (or in other domains) with advances in AI specifically related to government contracting - AI advancements requires effort and investment. Effort and investments usually follow the market - Industry is aware of the opportunity costs of their efforts and capital...what's the business case for pursuing AI in the government contracting field when there are likely considerably more lucrative opportunities available?

I think you’re correct that industry won’t focus on the federal contracting field for investing resources, at least at this stage.  From what I’ve heard and read about industry and AI development, government contracting will ultimately benefit from general purpose AI tools that are adapted to our field.

A good example of that is the original contract writing system developed 35 years ago.  Contracting users asked a series of multiple choice, yes/no, fill in the blank questions using a decision tree based process.  That “expert based” process wasn’t developed for the government market but government contractors saw a market for the tools after they were used for other customers.

Link to comment
Share on other sites

  • 3 weeks later...

I know this thread is old and I’m not sure what the original subject was.  But I saw something people might be interested in.  USDA is doing a demonstration of their bot contract close out process on April 13.  It’s open to federal employees and their support contractor.  You need to register in advance.  Details are in todays FAI News for the Acquisition Workforce

https://content.govdelivery.com/accounts/USFAI/bulletins/3503a2f

Link to comment
Share on other sites

On 3/9/2023 at 1:02 PM, Matthew Fleharty said:

All that said, I would love to be proven wrong by demonstrated capabilities, but most of what I've seen relative to AI applications to government contracting is either hope or hype.

Nic Challain the former Air Force Chief Software Officer recently created AskSage, a system built on top of OpenAIs GPT3.5, 4, etc. They've ingested acquisition.gov data up to FAC 2023-01 and other data sets. ChatGPT was limited to data up to 2021 so, recency was an issue. I messed around with it, pretty good. Most of the AI tools, including this one, function best as research assistants...or at least that's how I've used them.  Whether it's a net positive or negative, unsure. It's most definitely a technology we will be forced to deal with.

Link to comment
Share on other sites

I like using a bot to do things like contract close outs.  The process doesn’t involve much thinking and it’s mostly checking administrative compliance.  If it interfaces with financial systems, 1102s have little to do.  Agencies often ignore timely close outs and lose the ability to reprogram funding for other purposes.

Link to comment
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.

Guest
Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.


×
×
  • Create New...