Coping with Copilot

Earlier today, I received an e-mail from Microsoft Copilot. What grabbed me about the e-mail was not who it was from, but its subject line: "Built to do the impossible." When I opened the e-mail, the headline that jumped out at me was similarly bold: "A new AI era begins."

Wanting to learn more, I started surveying everything Microsoft has put online recently about Copilot. I was intrigued to find still more audacious claims, including:

• "A new era of innovation!"
• "A whole new way of working!"
• "It will unleash creativity, unlock productivity, and uplevel skills."
• "70% of people would delegate as much as possible to AI to lessen their workloads!"

Are your expectations set high enough yet?

The Promises Stage
Those of us who have been around long enough remember earlier Microsoft "big bets" -- Azure, Lync, SharePoint, BPOS, Dynamics and even "The Cloud" -- know that, eventually, Microsoft will spend enough to make sure that this particular bet lands well. We are now at what I like to refer to as "the promises stage." Microsoft is promising the sun, the moon and the stars -- and many, many others are joining in the chorus.

When you read more closely, it seems like the primary applications for Copilot are "search on steroids," plagiarism, e-mails written for you, and photos of you that are so nicely enhanced you might consider posting them on dating sites. You can also create illustrations of wild imaginings, or frighteningly real "deep fakes" showing people doing things they would never actually do.

Making matters worse, Microsoft finds itself besieged by an enormous early adopter user base that just wants more and more. Mobile phones took 16 years from introduction to get to 100 million users, plenty of time to figure out how best to use these new devices. The Internet took only seven years (thought exactly when the clock for this particular technology started is debatable), Facebook 4.5 years and generative AI...three months.

Three months, barely the blink of an eye in tech terms. People had plenty of time to figure out all those other advances, but for generative AI, they just went straight to liftoff.

What the World Needs Now: Practicality
The fact is that we in the channel all learned long ago what is required for the successful launch of any new technology: practical applications. What will the new thing do for us? How do we wrest value out of it? How will it improve our user experience, our employee experience, our customer experience?

The first thing I think the world needs to do right now is to chill out. Everybody from Bill Gates to Elon Musk to my barber is talking about how AI will eventually become superintelligent, and then it's, "Watch out, human race."

My response is really simple, and something we all learned long ago. Computers can do many things, but one thing they cannot do is generate a truly random number. By extension, it seems to me they won't be capable of random thoughts, either, or anything resembling true creativity that isn't sourced from something somebody else did that resembles what they've been asked to do. For AI, everything truly is derivative.

The other really great news is that there are some deep-thinking people out there who are actually producing valuable, practical applications for generative AI. Last summer, we presented a case study here in The Evolving MSP about CrushBank and how it leveraged its own 40 years in the IT channel, combined it with its extensive study of generative AI and IBM watsonx, and produced a technical assistant that could reduce the time it takes to resolve an IT problem by rapidly finding documentation, instructions and whatever else a tech needed to effect repairs. The company also devised a virtual help desk agent that answers user calls and provides useful advice. Practical on all fronts!

Educators are finding wonderful ways to apply generative AI to curriculum and materials development, lesson planning and even interactive activities. Doctors are receiving help from robotic assistants that can quickly retrieve patient details, display previous scans and more. Even the folks in sales and marketing are saving time and accelerating sales using recommender engines that compare their company's social media-augmented database of insights and information about their customers with a highly detailed database of all their product and services offerings. The AI figures out which products or services would be most desirable to each customer and sends them the appropriate marketing material. It then informs the appropriate salesperson for each customer about what was sent, enabling them to quickly follow up.

Of Adoption and Acceptance
Every software development project is best measured by its rate and degree of adoption. Three months to 100 million users is a definite vote in favor of the "success" of AI. By engaging in such massive, overt hyperbole, Microsoft is challenging its own customers to take it seriously in its approach to Copilot. It would be all too easy for this "big bet" to fall flat on its face, which could indeed be an existential event.

For comfort, let's close with the wisdom of Geoffrey Hinton, the "godfather of AI," who said, "In science, you can say things that seem crazy, but in the long run, they can turn out to be right. We can get really good evidence, and in the end, the community will come around."

(It's worth noting, though, that Hinton also quoted the Rev. Martin Luther King Jr., who said, "Our scientific power has outrun our spiritual power. We have guided missiles and misguided men.")

Posted by Howard M. Cohen on June 20, 2024