Some of you think that AIs are going to eat the world. Others think it’s just for hallucinating events that didn’t happen, or making cat videos. The truth is that the latest models hallucinate much less, it was important to fact check people, and the best AIs are smarter than most (nearly all). That’s happened quickly, and so the question isn’t ‘what are we doing with them today’ but ‘how will they remake the world over the next few years’.
All of the billions spent on AI innovation just so we can get these beautiful cat videos pic.twitter.com/HKHReIuscy
— Bindu Reddy (@bindureddy) March 8, 2025
I’ve written that I probably wouldn’t have AI replace co-pilots on commercial planes, although soon they will clearly outperform humans, and pushing back against AI co-pilots will compromise safety.
Already people plan their travel mostly online, doing it themselves, because all but the priciest trips lack the margin to have the supplier pay a real travel agent. Soon AI will do a much job planning trips, which is why Expedia is laying off people but spending heavily on AI.
But that all misses the forest for the trees. AI seemingly knows everything because it trains on everything. That makes knowledge of facts far less important. It’s a better technical writer than you are (and I am). What it isn’t is human and that makes what’s distinctively human still valuable and more important than ever.
Here’s a piece of a longer discussion that I think makes an important point about the future of travel, without doing so explicitly. Travel is going to matter more than ever because of AI because informal knowledge and social networks will take on greater importance.
- The AI knows everything that isn’t secret.
- Your knowledge of public information will approach zero value.
- What isn’t published, then, becomes relatively more unique and valuable.
‘
- Trading secrets’ – getting out and having discussions with people, trading off-the-record information and gossip becomes more valuable.
The internet, and zoom, connected people online and in some sense reduced the need to travel. Many business meetings that were once in-person now happen over Teams. Your job might even mandate Return To Office but you find that you still log into meetings where other employees are back in the office, too.
But AI will increase the returns to social networks. Some of that happens through online communities, Telegram chats, WhatsApp… but often those relationships are still formed and nurtured in-person.
AI effectively increases the relative return to and importance of in-person and therefore travel. (And AI will help us plan and get more out of travel, too.)
[I]f you want to get things done, you’ll need to mobilize resources.. you’ll need humans.. your network of humans is not 20% more valuable, it could be 50x more valuable.
The most effective people will have an army of AIs at their disposal so are that much more effective. It’s going to be far more than Delta using AI to extract more money from passengers.
I do think AI will know more secrets than this discussion gives it credit for, remember Google tracks all the movements on Android phones, stores your email, AIs know what you ask it and can surmise why. But instead of ‘no need to travel because the AI tells you anything you’d learn’, forming personal relationships and gaining special insight will give you information you can’t get from AI and gain context that will help you effectively prompt, understand, and filter what AI tells you.
My father was a road warrior long before the term became popular. Dad was traveling 18-20 weeks/year, much of it by air. He became a member of United’s 100,000 Mile Club soon after the United/Capital Airlines merger in 1961.
Dad always believed that the secret to his success as a salesman was that he called on customers in person. “It’s much harder to say ‘no’ face to face than over the phone. They can always just hang up the phone. And, in person, I can find out the problems with the use of the product (bulk cotton in carload lots), meet their objections to purchase and ask for the order.”
AI can’t – and never will – replace human interaction. It will always be easier to say “no” to a robot.
At its current stage, artificial intelligence software in language models, like Chat GPT, and in visual generators, like Dall-E or Midjourney, etc. are merely ‘tools’–which can also be ‘weapons’–but it is not yet sentient or uncontrollable, like Gary mocked with his first line in this post, ‘eat the world.’
Since we, humans, are practically engaged in an arms race to advance AI to the next level, verses competitors and global rivals, we really are missing the mark on safe guards and governance–this is usually how things end up going awry and hurting people. I am concerned that we’re going to have too many folks out of work too fast based on the false promises of a panacea with this technology.
There is a real probability that we’re offshoring and automating at the rate that many of us in USA are simply not going to have any ‘real’ work to do anymore. If we get robotics up to that speed as well, then we won’t even need many manual labor ‘jobs’ either. If our governments don’t prepare for this, it’ll be chaos and war that follows. People need a purpose and something to do with their time. Escapism can only do so much for us. Food for thought. Yum.
It’s funny that even the advertisements for AI can’t even pitch anything interesting. Billions and lakes burned to recreate….maps. Well done! I don’t think Salesforce even knows wtf to do with AI.
My friend and mentor used to say “Business Development is a contact sport.”
Many in the business world currently think in only current cost terms. Save money, cut travel and sales staff.
The immediate loss is the person to person network which doesn’t rebuild quickly.
Also lost is the trust in continuity.
Business travel is always seen as a high cost questionable benefit by the finance types.
200 days a year for 20 years proved very effective for the companies I worked for.
Gary Leff writes, “The AI knows everything that isn’t secret.
Your knowledge of public information will approach zero value.
What isn’t published, then, becomes relatively more unique and valuable.”
To keep our AAdvantage miles unique and valuable, no one should ever publish the secrets of redeeming AAdvantage miles.
Counterpoint: the LLMs out there have already trained on the available texts, copyright be damned. That’s produced some fancy models that generate responses that would get you a C+ in college. That makes sense: C+ is an average score, and if you feed AI data from average performers, you’re not gonna do better.
So the latest generation of models uses some fancy tricks to appear smarter. At the same time, groups are trying to enforce copyright, to prevent LLMs from stealing their lunch. And others are using LLMs to spew content to the internet.
That means that nobody’s gonna be able to train the next generation model on information available on the internet. If it’s not legally blocked, it’ll be full of LLM-generated content. So these companies are searching for pristine content, which they can’t steal anymore. And nobody’s showing a profit.
My biggest fear is that AI will never be as useful as it is right now.
You know, you remember when Google search gave you what you were looking for, immediately?