When the U.S. president was impeached last week, it surfaced the fact that impeachment, and particularly the removal of a U.S. president, is a process that doesn’t work. Throughout U.S. history, there have been 45 presidents, and given that no one trains for that job, you’d figure several would have been removed for cause. However, impeachment has been attempted three times without a removal.
If a process that has been in place for more than 200 years has never been successful, it is a really bad process. So, rather than focus on whether President Trump should be impeached, I’m going to suggest a technology-driven process that should be both fairer and more effective.
I also think arguing about the impeachment process rather than President Trump’s impeachment itself will be a far safer way to spend the holidays.
I’ll close with my final product of the week: a new Nvidia offering that could make Christmas shopping massively easier, have a better outcome, and allow companies like Motorola to compete more effectively.
Removal for Cause
Every company has a process to remove an employee for cause — from entry-level to CEO. Certainly, you could argue that many of those processes aren’t fair. Though unacceptable, it is far from uncommon that an employee who rightfully complains about unsafe conditions, or about a superior’s harassment, abuse of power, or inappropriate or illegal actions, is still likely to be terminated.
It’s also likely that a CEO who is terminated will get a huge golden parachute, while a terminated rank-and-file employee often is lucky to get cab fare home. In the gig economy, it isn’t even clear whether a worker is an employee in the first place.
On this last point, I’ve seen abusive treatment of contributors by YouTube and Amazon, and they don’t seem to understand they are screwing with people’s livelihoods, which often becomes a foundation for unionization. If that happens, both firms will learn the wonders of collective bargaining, and that is generally the opposite of fun.
HR, which was supposed to be an advocate for the employee and a less aggravating alternative to unions, at least in tech companies, evolved instead to become a tool of management, and that is one of the reasons we see signs of unionization at places like Google.
I’m sure we’ve all had managers who sucked and put our job at risk. I had a guy who didn’t give me my annual objectives until he did my annual review, and then had the nerve to give me my first bad review because I didn’t meet objectives I’d never seen.
In that case, I successfully escalated but then had to transfer out of the department to get away from extended retaliation. Eventually I left the company because using HR effectively got me a reputation for being a “problem” employee.
Like many, you may work for a company that has an “employment at will” doctrine that allows the firm to terminate you for any reason (likely another cause for what seems to be an upswing in union activity). In a tight labor market, that policy likely is hurting qualified employee acquisition, particularly in areas like data scientist, where there are far more openings than people.
Oh, and covering up problems that employees identify hasn’t exactly been a brilliant strategy in or out of government, because social media is a thing, and those stories now have the potential to become national or even worldwide scandals. These scandals damage the firm’s brand, reduce its competitiveness, harm revenues and valuation, and put the firms’ boards in bad spotlights.
As a result, several companies are developing AI-driven programs that more fairly manage this process, better ensuring that employees who are assets are cared for properly, that problems are discovered and dealt with in a timely way, and that problem employees (and managers) either are retrained or removed expeditiously.
For CEOs, there is a very different process. The board of directors must vote out the CEO, and given that CEOs often pick their boards, and too often are the executive chairman (who sets the agenda), this can be problematic.
This process is one of the reasons we have CEOs in place who are famous for behavior that would get any manager fired, and why those CEOs can do incredible amounts of damage before being removed. Still, even after damaging their firms, they often get millions in severance pay, which generally sits poorly with employees, customers — and particularly, investors.
Still, we do remove CEOs, and we have yet to figure out how to remove a president.
The Impeachment Process
Impeachment works similarly to the grand jury process. The House of Representatives (like the House of Commons), represents the people of the U.S. (think popular vote), to prosecute the case.
The problem is that if the opposing party has a majority, it will look like a partisan attack, regardless of evidence, which makes the president’s party feel unfairly treated. If the president’s party holds the majority, impeachment is unlikely because it would substantively reduce its power, and we are a power- and influence-driven race.
Once impeachment occurs, which is similar to a prosecutor filing charges after a grand jury hearing, the process then moves to the U.S. Senate (like the House of Lords), which represents the states (think Electoral Collage). Then the Senate holds a trial presided over by a Supreme Court justice who has been nerfed (has a fraction of the procedural power of a regular judge).
The Senate acts as a jury of sorts. Removal requires a vote of two-thirds of the senators. This percentage is an excessive threshold; it was not achieved in the two previous attempts, and it is unlikely to be achieved this time.
Again, assuming votes along party lines, the process appears corrupt. The senators in the president’s party are influenced by the reduction to their power that the removal would cause, and those in the opposing party appear corrupted by the desire to get more power.
In short, the process not only is inherently unfair, but also ineffective. Instead of preventing a civil war or a misbehaving president from staying in power, it seems to ensure one of those outcomes (fortunately more the latter than the former).
The AI Blended Fix
First, identification is what AIs increasingly do well, and there have been massive advancements in speech-to-text. Nvidia announced one such advancement at its Asian GTC Conference last week.
Most everything a president does is documented. The AI, using machine learning, would be trained to look for bad behavior. If it hears something that crosses the line, it first alerts the president and his core advisors (based on presidential preference) so they can avoid making a mistake in the first place.
In short, the AI’s first and foremost job is to function as a training tool for the president and his immediate staff, to ensure they learn the lines they can not cross and don’t cross them.
If enough evidence of bad behavior is collected, the AI issues a recommendation, which may include additional training, censure or removal from office, depending on the problem and frequency.
That recommendation bypasses Congress, which has shown itself to be too biased to execute this process. The full Supreme Court (which serves for life and doesn’t have the reelection and power problems Congress has) decides on the outcome, which could be training, censure, or removal from office. There might be a fourth or fifth choice, which could include fines and incarceration, depending on the offense.
Wrapping Up: Creating Fairness
Whether we are talking employees, CEOs or U.S. presidents, the process surrounding the termination of any of these folks isn’t optimized or fair. Regarding impeachment, this is particularly true, given how biased Congress is, because as a body it is motivated to be more loyal to party than to country, corrupting the process.
All of these groups need better oversight and warnings to avoid problems in the first place and a fairer process when the termination becomes an option. A properly trained AI certainly could perform the duty of warning and fairly starting the remediation process.
The Supreme Court is the ideal body to finish the process for the U.S. president, but there is no equivalent body in a company, where the people making the decision wouldn’t be subordinate to the CEO whom they might have to remove at some point.
The closest might be internal audit, which typically reports through the CFO to the board, but that would require changes. Right now, internal audit effectively is subordinate to both the CFO (who reports to the CEO) as well as the board, (which also may be subordinate to the CEO).
Putting your career at risk to do a job is a conflict that should be avoided. This conflict suggests that internal audit also should report to an outside organization that can protect its independence.
Another idea would be to create a parallel organization of life appointees to the Supreme Court, which would have investigative responsibility and could build a case against anyone in any of the three branches of government.
Staffed with top career investigators, this could provide a better, unbiased approach to removing anyone from office. They could be assisted by an AI that would allow scale without requiring another massive government agency.
By the way, talking about how to fix the impeachment process might be a better discussion this week (as in less likely to cause you to leave a family gathering early) than whether the president should be fired — just saying.
Nvidia launched several interesting things in Asia last week, including improvements to its self-driving platform, that real-time conversational AI I mentioned above, and Orin, which it represented to be the world-leading robotic processor (after watching the latest Terminator movie, I promise to say only nice things about it. The big announcement for me, given the time of year, was its new recommendation engine based on its AI platform.
This engine is having a massive positive impact on both Baidu and Alibaba, reporting up to a 10x improvement in click-through and a 100x improvement in throughput. More importantly, it appears to be improving customer satisfaction as well, because customers are finding products that better meet their unique needs.
Given that online retailers like Amazon and social networks like Facebook are capturing tons of information about our loved ones and us anyway, why not use that information to make better recommendations?
Done right, they would know not only what we like, but also the configuration we prefer. This process could be particularly useful in picking a car, and it could make the broad product lines in companies like Motorola rise to their full potential by allowing the firm to recommend the phone and configuration that would be the best for a prospective customer.
This result should overcome the advantage held by simplified lines, like the iPhone, in terms of making a choice easier and removing friction from the sales process.
With this tool in place, a big vendor like Amazon could provide a list of recommendations tied to every person you regularly buy for, based on the budget you’ve provided, tied to what it knows about what each of those people like.
You’d then make your selection for each or just hit “accept,” and the packages would be wrapped and sent automatically (the AI even could recommend what goes on the card in the box).
The process could start when the Black Friday sales start and optimize for those products that have deep discounts getting you the most bang for the buck. If it were part of a service like Prime, more people would sign up.
I think this recommendation engine from Nvidia could change retail, and so it is my final product of the week for 2019. Next week I’ll name my product of the year and product of the decade. You may be surprised by what makes that cut.