None of Us is as Dumb as All of Us! Unless, Of Course, You Include AI!

the doctor sees a lot of unsolicited pitches hit his inbox each and every day. Since SI does not cover press releases (since he just does not give a damn about your meaningless marketing sound-bites which do little to nothing to advanced education and technology) he ignores most of them. But this week he saw one of the most ridiculous headlines ever:

Can AI Harness the World’s 2 Billion Social Media Influencers?

Ignoring the fact that that the headline is factually incorrect (there are 2 Billion Users, NOT Influencers), this is one of the dumbest questions ever posed and anyone who understood anything about the state of AI today would not even want to ask it!

Generally, if you are going to train AI, you want to train AI on expertise. And where’s the last space you’d expect to find expertise? That’s right! Social Media.

But that’s just the tip of the why-you-should-not-do-this iceberg! If you include everyone, you not only include everyone of above average intelligence, but everyone of below average intelligence by very definition. So while you will have a few geniuses, you will also have morons, imbeciles, and possibly even idiots (as per the original Binet IQ scale). Do you really want them training your AI?

Moreover, what do people share on Social Media? Their most brilliant ideas? Well developed arguments? Philosophical contributions? Wisdom? Profound insights? Or pictures. Comments on politics. Viewpoints on pop culture. Their thoughts of the moment. Complaints. Rants. Digs. Manifestos. Insults. Self Praise. And so on. And most of it in blurbs, not sentences, and definitely not paragraphs. And in addition to the onslaught of bad grammar, the rate of spelling errors is atrocious.

Is this what you want to train an AI on? Really? You really want an AI that is going to make decisions like an angry dumb, self-obsessed, neurotic, troll with self-esteem issues making your decisions? And that’s likely a best-case scenario.

the doctor doesn’t know about you, but if he’s going to trust an AI, he wants that AI trained by experts for specific tasks, with performance analyzed and tweaked by other experts, since, as we know, there is no such thing as artificial intelligence, since no algorithm is intelligent, no matter how advanced, and what we really have are advanced automated reasoning algorithms. But if those algorithms were trained by the impaired, those algorithms are the last algorithms he would ever want to use. And those algorithms should NOT be on your list either.

The implications of Crying Thief!

Today’s guest post is from Tony Bridger of Assymetrix Consulting. Got a spending, process, or change management problem? Tony has a solution.

There is an old Nigerian Proverb that runs a little like: “One cry of “Thief!” and the whole marketplace is on the lookout.

However, crying “thief” has serious implications for many business, particularly those public organisations with shareholders who would quickly perceive financial crime as a systemic business process failure.     It is easier for management teams to internally manage fraud than to prosecute. Detection of large fraud is also an admission that both controls and deterrence are failing.   In a recent article, It’s Hard to Find Fraud in Big Spend Stacks …   the advent of AI could provide that vital detection of internal fraud.   It’s a sophisticated solution.

Whilst we are on the subject of proverbs, a key element in fraud management is “prevention is better than cure”. Companies that detect fraud have clearly not created the cultural norms that others take for granted that deter staff from committing fraud.   There are many cultural and technological capabilities that can reduce the incidence of fraudulent activity that are well within the grasp of many businesses.   Deterrence – or risk of detection is a critical cultural message.

With some careful risk analysis, it is quite easy to map out where company fraud is likely to originate. Finance, Procurement and staff expenses are usually the key internal risk areas.   Culturally, one of the first steps is to ensure that there is adequate separation of duties.   In finance, this is simply ensuring that a finance staff member does not have the capacity to both create a supplier vendor master entry – and pay an invoice.   This is a system administration role setting. The creation of “dummy vendors” and subsequent payments is often down to this simple failure.   Making all data elements (Business Number, address, contact details) as mandatory data items also reinforces the message on data integrity.   Many mid to high end systems will also allow user audit trail analysis if required. This simply captures the user-id of the employee accessing the key finance system forms.

For smaller companies, separation of duties can be an issue – but keeping a register of new supplier entries and reviewing this regularly is a key move.   In the procurement space, the person who creates the contract and then manages the winning vendor should also not be one and the same person if possible.   Again, hard to mobilize with limited staff and expertise – but a very clear signal around why is a powerful deterrent.   The idea is not to create a draconian working environment – it is simply ensuring that employees understand that this is designed to protect them – as well as the company.

Where possible, organizations should also use the power of their accounting system to the full.   Many of the low-end accounting systems have decent quality automation for transactions like staff expenses.   From experience, there are some subtle employee mindset changes generated with increased automation.   Almost all of us realize that entering data in to a system creates a record.   Once submitted, unless a request is made to vary the claim – the electronic evidence exists.   Paper can be lost, shredded or misinterpreted.

Almost all staff will recognize that these transactions can be retrieved many years later.   A very good business practice is to engage a vendor that provides duplicate invoice analysis services periodically.   This service can also detect anomalies and “odd” transactions.   A multiple repeated “same value” claim by an employee will almost certainly be found and analyzed. As many of these services are contingent based, they are quite affordable.   Regular auditing can also send clear signals on fraud risk assurance.

However, the combination of separation of duties, increased electronic transaction processing and periodic data analysis should send very clear cultural signals about what is acceptable. Staff will work out the “why?” comparatively quickly.

Organizations cannot effectively function if trust is lacking.   The notion of the cry of thief! Is far more acceptable if good management controls are in place and any subsequent fraud is detected. In effect, it’s a best effort approach to fraud prevention.

Thanks, Tony.

Why You Have to Find that Fraud in Big Spend Stacks …

We recently published a piece on how it’s hard to find fraud in big spend stacks, and it is an important one. While fraud in most organizations might be relatively small, and might be mostly controllable by the right culture, processes, and systems (but that’s a subject for a future post), it’s still going to be there, and the most common form of fraud you are not going to detect is collusion fraud.

But this can be the most costly. Let’s say Bill and Ted both have invoice approval rights in the services procurement system and can singlehandedly approve services procurements up to 20K. Let’s say Bill’s buddy Bob has a services firm and let’s say Ted’s buddy Tim also has a services firm. Let’s also say that the organization also has a great need for temporary contingent labour to man the warehouse, clean the offices, and guard the assets of the company.

Let’s say that oversight of these services is left up to the approver for verification. Let’s say that Tim routinely sends two services guards when the general policy is to have three guards on duty and that Bob typically sends only two janitors to do the work that would typically be done by four by the old services provider. Who’s to say that Tim doesn’t send two guards but bill for three? And who’s to say that Bob doesn’t send two janitors and bill for four? And if these invoices are sent bi-weekly, they are going to fall well within approval limits.

Moreover, who’s to say that Ted doesn’t know about Tim’s over-billing and Bill doesn’t know about Bob’s over-billing? And who’s to say that Bill and Ted don’t have a deal to approve the over-billings for each other because their wives are getting an “efficiency consulting” fee from Tim and Bob’s companies?

Maybe this doesn’t happen in your company, but it happens more than one thinks, and just because you never detected this, how do you know it’s not happening? Invoices from real suppliers for real services at approved rates can still contain fraudulent over-billings for services not actually delivered, and those proceeds can still be partially kicked back through indirect channels to organizational employees.

But how do you detect this? Very sophisticated AI-based algorithms that detect unusually high approval patterns between two organizational employees, for amounts that should have been reduced with new contracts, that don’t match typical, anonymized, organizational patterns. And then human investigation to find the truth.

So why is this so important? Besides plugging the leaks? Because if you can’t find internal collusion, how will you ever detect potential cases of external collusion? And gather enough corroborating evidence to at least get an investigation going? If industries collude, and jack prices above market prices, the organization will lose considerably more than it will lose to Bill and Ted (from the evil, parallel, universe). And this happens more than you think too, it just doesn’t always get detected and investigated. Fortunately, sometimes it does, and sometimes, even if there is no certainty that fraud happens, regulators, presented with enough evidence still investigate — like they are doing now among the German automakers (which led to a surprise raid on BMW headquarters as recently reported in the New York Times) that are suspected of conspiring to hold down the prices of crucial technology (as initially reported in July). Regardless of the outcome, technology that can identify potential fraud and gather correlating evidence will keep everyone more honest, and that’s a good thing.

Are You Doing Your Own Quality Spot Checks? And Should You Be?

By now, if you haven’t heard of the Kobe Steel Scandal, you’ve been living in a cave. (Which, in some organizations, is highly probably given that one of the tricks the CFO likes to do to Procurement when fiscal year end is approaching is to lock them in the basement until the mandatory savings objective is reached … hence our post yesterday on why every day is Halloween for some Procurement departments.).

This scandal is scary. Not only because the data falsification on strength could go back as far as 10 years on some batches, and who knows what bridges, high-rises, and busses that steel has gotten into (and even a .1 degradation, while not enough to jeopardize immediate safety, can impact expected life span and increase susceptibility to decay, making safety a concern down the road before inspection and maintenance schedules kick in).

But this brings up a good point? If more companies were doing more spot checks on shipped product and quality, instead of just trusting Kobe, would it have been 10 years before the scandal was exposed. Even if only a small percent of batches are affected, I highly doubt this would have been undetected for 10 years, even if only one bar or sheet in multiple shipments were tested.

This is an example of what happens when finance tries to get too greed or supply chains to lean by centralizing a function downstream. When one party is responsible for everything, one failure can reverberate up multiple chains undetected — and have potentially disasterous consequences. Now one might say this problem is solved by co-locating people on-site, but if those people never leave the site, even though you pay their salary, their work family is the people they work with day in and ay out and the existence of that company is their livelihood. Are you sure they won’t bow into the local culture and, if the culture dictates, defer to authority or collectively hide the shame?

Just like third party audits are needed, for critical materials, so are third party quality tests. Doesn’t have to be you, could be an independent organization set up between your co-opetition that does random independent quality spot-checks on 1 in 10 shipments and shares the data with everyone.

Just like a good Chef would never use an ingredient without insuring it’s quality, a good Procurement organization should never let a shipment be accepted without a high degree of confidence that it’s a quality shipment. And confidence like that only comes from organizational testing or trusted third-party independent testing. So don’t get too lean or too cheap — your organization, and the lives of its customers, could depend on it.

For Some Procurement Departments, Every Day Is Halloween!

A week from tomorrow is Halloween. While until recently meant to celebrate All Hallows’ Day, it’s common observance today is to provide an opportunity for kids to trck-or-treat, high school and college students to throw parties, and adults who miss their childhood an opportunity to play dress-up once again.

As part of this festival of tricks and treats, celebrants (regardless of religion), will carve pumpkins into jack-o’-lanterns, light bonfires, bob for apples, attempt to divine (and contact the dead), play pranks, purposefully visited haunted attractions (whether supposed to be fake or real), tell scary stories, and even watch horror films.

But for some Procurement departments, this is every day. Every day they are beaten up by the CFO and feel like they are being carved up like jack-o’-lanterns as their performance is dissected with biology lab precision. (Performance that’s not as good as it should be since the CFO won’t let them buy the best tools.)

Stakeholders, who bring (new) requirements, are constantly lighting fires under the team at the last minute, not realizing that great results takes great planning, and that doesn’t happen overnight. That sometimes strategic acquisitions take 3 to 6 months of hard work to find and unlock the hidden value, and that one cannot expect miracles when Procurement, already at 100%, is asked to (re)source a category 30 days before contract termination (and the date was known 1065 days in advance when the last contract was signed without Procurement involvement.

Senior buyers are constantly bobbing for new opportunities, checking out random whims because they don’t have a modern spend analysis system to help them identify the best possibilities for savings. At some point they will get so desperate they will go to gypsy diviners in their quest to identify savings opportunities, and even ask to speak to dead business gurus of ages past.

They will feel like they are constantly being pranked by sales people as they won’t have the insight to build true-cost models, the IT dungeons they have to go to for help normalizing market data from the free feeds they have access to will feel like haunted houses, every new request made from them is a scary request, and their entire
existence feels like a horror movie.

And the scariest part of this story, is that it’s not just a few departments, it’s a sizeable number of Procurement departments. Remember, 40% don’t have modern Supply Management tools, and of those that do, the majority have major holes in the Source to Pay to Delivery Cycle.

So, CFO, this Halloween, only you can change Procurement’s existence by giving them a treat — the budget to acquire new systems next fiscal year. Trust me, Procurement needs them.