Read it again...

Due to the overwhelmingly positive response and engagement we received on Carly Fiorina's previous newsletter post about AI, we have decided to rerun this week. This topic is truly significant, and the depth of Carly's insights resonated with many of you. As we approach the upcoming 4th of July long weekend, we believe it is the perfect opportunity to revisit this thought-provoking piece. We understand that this post is longer than most, but we firmly believe it deserves another round of attention and discussion. Thank you for your continued support and enthusiasm. We look forward to engaging with you once again on this important subject.

-Team Carly

As a graduate student at MIT, I took a course on Neural Networking. I was earning an MS degree in Business, not Computer Science or Engineering, so this was a di?cult subject for me. Nevertheless, I thought it was important to try and understand a technological breakthrough that allowed computers to react, respond, and “learn” through a web of “synapses” and connections rather than a hierarchy of commands and controls. It did not take much technical training to recognize that there could be huge implications.

Fast forward and AI has arrived. It is the subject of TED talks, congressional hearings, regulatory reviews, and ubiquitous media coverage and is urgently discussed in every board room. Every industry is evaluating how AI can help generate revenue and lower cost. Small companies and countries see the opportunity to overcome their scale disadvantages. Investment capital is pouring into the field all over the world. At the same time, increasingly dire warnings that AI poses an “existential threat” to humanity, as well as urgent pleas to pause, evaluate and regulate, are being issued by many in the technology community.

If we are clear-eyed about AI, we probably conclude the following:

1. There will be no “pause” or a slow-down, even if there should be.?Too much money, entrepreneurial energy, and innovation are now focused on AI as the very, very big “next, big thing.” Many in the field have said this is the fastest technology curve they have ever experienced.
2. There are real perils, and we will not understand all of them in time to prevent them.?Just think about how social media has evolved. We are still figuring out the many negative impacts on human behavior and mental health. We still do not know how to prevent misinformation, disinformation, fraud, division, and discord. Authentication and trust have become more, not less, di?cult over time. Now imagine the impact of “Deep Fakes,” virtually indistinguishable from reality, on all this.
3. There is a real promise with this technology: the promise of more engaging education, more complex problems solved, vastly superior customer service, mesmerizing entertainment, meaningful connection, and community, not to mention improved profitability. But as with every promise that comes from innovation, the dislocations will be more profound than any we have seen before.

In the face of this, how should leaders respond? What is the best way forward in an environment that feels increasingly precarious and truly unprecedented?

1. You can start with humility.?Many industries and executive suites are typified by hubris. So are many corridors of government. “We will figure this out.” “We know the answer.” “You don’t understand the complexity and sophistication of all this, so just leave us alone to do what we do best.” These are the sometimes too common attitudes of people who believe they are the best at what they do. AI should humble us all. Now, it is the wiser course to admit: “We don’t know.” “We are struggling to figure this out.” “We need to work together.” This sounds so basic, and it is. Nevertheless, adopting an attitude of humility is profoundly di?cult when hubris, or at least supreme confidence, has been rewarded for so long.
2. Proceed with clear-eyed realism, not over-confidence or hype.?We should be transparent and candid about what we know and especially what we don’t know, about what we can and cannot do. Every assessment of the upsides should be balanced by evaluating the downsides. No one should “hide the ball” about what can go wrong in any application of AI. Nonetheless, no one can a?ord to sit the game out on the sidelines either.
3. Make the time, and take the time, to deliberate thoughtfully and carefully.?This will be increasingly di?cult because the pressure to “hurry up and move” intensifies daily. Nevertheless, in a highly complex situation, where the possible pitfalls and missteps of AI can be as ruinous to an enterprise as the upsides can be transformative, a sound decision-making process is paramount. Which decisions are timely? Which should we wait for, other events or more data? Which options are low-risk, and which are currently high-risk? Are some choices dependent on others, thus suggesting an ordering of actions and priorities? Good decisions come from deliberate, careful, thoughtful decision-making processes. Such a process requires su?cient information, analysis, discussion - and time.
4. Ask lots and lots of questions.?The right questions have never been more important. In most organizations, having the right answer is valued and so people strive to be the one with that right answer. Now, with so much unknown and the breakneck pace of change, the right questions are the best tools we have. There will be no one right answer. There will be many possible answers, and they will change over time. Figuring out the pros and cons and the “so what” of all those possibilities will require acute questioning abilities. Asking questions is also an e?ective safeguard against groupthink. Groups of very smart people frequently drive themselves o? a cli? with a really bad decision because, at some point, everyone stops asking questions - even seemingly obvious questions. Groupthink happens most often when people feel under pressure: the pressure of time, the pressure to conform, the pressure to impress, the pressure of scrutiny and criticism. AI creates all those pressures, so the danger of groupthink is heightened. Ask questions.
5. Identify problems early and often.?Don’t ignore them, don’t bury them, and reward those who surface them. Admit mistakes and course correct.AI promises to be a very bumpy ride. Problems will abound. Spotting and raising awareness of them, at a micro or a macro level, is the only way to mitigate or solve them. If a mistake has been made, a miscalculation has occurred, own it and fix it. Of course, this also sounds so elementary; why am I bothering to highlight it? Because in many organizations, problem identification isn’t rewarded, particularly when the solution isn’t obvious. If you’re in an industry where a hyped-up answer is the best way to get ahead or?pop the stock, coming forward with a problem and a humble “I don’t know the answer” requires a big shift in mindset. So does an admission of error.
6. Don’t stay in your silo. Collaborate constantly.?Seek out di?erent perspectives and expertise. For all its promise, AI creates lots of Fear, Uncertainty and Doubt. Normally our instinct is to hunker down when the FUD factor is high. Now, however, collaboration is vital because no one knows enough to be smart enough on their own. As an obvious example, technologists now need regulators, and regulators need innovators. One industry can learn from another. Players within a field can learn from each other. Parents can learn from others in their community. So can educators at many levels and in di?erent disciplines. Part of the great value of these kinds of collaborations is the harnessing of di?erent points of view. A history teacher will have a di?erent standard for authentication than a social media company. A regulator’s concerns for safety and the public good help balance the push for growth and profit. What makes the most sense from an engineering perspective might not work for a customer. Although they have flattened over time, organizations are built around hierarchy, silos are strong, turf is protected, and collaboration across boundaries is di?cult. Collaboration between separate organizations, or even separate industries or fields, is exceptionally di?cult. And yet, that collaboration is required now. Fortunately, there appears to be a great appetite for such collaboration. People seem eager to reach across boundaries and learn from one another precisely because Fear, Uncertainty, and Doubt are very real and omnipresent. We are smart enough to know that AI is bigger than each of us and our organizations. To capture its promise and manage the perils as e?ectively as possible, we need collective conversation, reflection, learning, and action. We all need to become more like neural networks - connecting in unexpected and new ways.
7. Trust is harder to earn and more important than ever.?We already live in a world where it’s hard to know what’s true, what’s real, and what’s authentic. AI will compound this problem exponentially. The worst of AI is capable of “infecting” even the most trusted of sources and brands. The internet is awash with garbage - the fake, the false, and the fraudulent.?AI “learns” from the internet, and as the old saying goes: “garbage in, garbage out." There is a great deal of research being done on why AI “hallucinates,” on how to increase the probability that real and true content can beat out what is fake and false, and on how to erect “guardrails” against the most dangerous of AI’s tendencies. I am personally doubtful that the research can keep pace with the development of AI’s capabilities, especially because investment dollars dwarf research budgets. Trust always matters. It matters more now. Trust isn’t earned by being perfect - perfection is never possible, mistakes will be made, and bad things will happen. Trust is built and preserved when leaders remember to: act with humility, not hubris; stay clear-eyed, candid, and realistic; deliberate carefully; ask questions rather than pretend to have all the answers; identify problems, and admit mistakes; and willingly share, collaborate and learn from others.