Who wants a better chat bot?
Jul 24, 2024 12:31 pm
Hi there,
My summer break is almost there! Before I leave, I'll travel to South Africa to speak at the HR + L&D Innovation & Tech Fest about how we can keep work uniquely human and about future-proofing payroll! Will I see you there?
I'll be talking about Earned Wage Access in this webinar on July 24 organized by PayrollOrg and DailyPay, Inc.
Who wants a better chat bot?
When you ask people if they want a better chat bot, they almost always say yes. But when you dig a little deeper, what they really want is better service. While everyone likes the instant access we get through apps, when things get tough, we want to speak to a person. Just look at people complaining on social media about customer service issues: everyone has a story about a simple problem that became a huge problem because they couldn't talk to a service agent, a person. It's a real issue.
For nearly two years, we've been hearing that generative artificial intelligence is improving every day. That it will replace people. That there won't be any work left, and we'll live on Universal Basic Income. I've always doubted that idea.
When I hear that a single technology is the ultimate solution to every problem, I take that with a grain of salt. I know we have to get through the hype, before we find out what really works.
But also, I had not kept up with AI developments the way I should have. And when I started to study, I was a bit surprised to discover that the tone about AI is changing. Some critical observations are coming from unexpected places, which makes me wonder:
Is the AI bubble (finally) ready to burst?
And by burst, I don't mean disappearing like the metaverse and blockchain. I mean becoming more grounded, where we see AI as a useful technology rather than the solution to every single problem.
When everyone is in awe of something new, no one is asking the hard questions anymore. And whenever someone questions AI, the whole tech community jumps on them, insisting they don't get it. That they are wrong.
But we need people who hold up a mirror to show us what we're doing and ask us to reflect on that. We need people who critically examine AI, because no technology is purely positive.
So, when I found the Ludicity blog, it felt like a breath of fresh air. The author manages to bring the hype back to its proportions. He states that:
"unless you are one of a tiny handful of businesses who know exactly what they're going to use AI for, you do not need AI for anything - or rather, you do not need to do anything to reap the benefits. AI, as it exists and is useful now, is probably already baked into your businesses software supply chain."
I hope that quote makes you curious enough to read the article. But also a warning: the language is colorful, so if that offends you, do not click this [link]!
As you know, I enjoy listening to the Bubble Trouble podcast and this conversation with Andrew Orlowski was enlightening. He wants AI to be useful. But the problem is the same from blockchain to sharing to crypto to the metaverse: we haven't seen the expected productivity improvements from these technologies.
And now, he explains, expectations for AI have been raised too high. Consider this observation: "It's the lazy and dishonest people who benefit most from AI. And they are going to be using it all day to pretend they are busy." You definitely want to listen to this interview!
Look who's starting to wonder about AI...
I have to say that I was a bit surprised to notice that even investors, thought leaders and advisors are having a more measured debate on AI (I won't call it second thoughts yet). And when they are finally starting to ask hard questions, we better take notice!
Goldman Sachs reports that despite an estimated $1tn investment, there’s little to show:
But eighteen months after the introduction of generative AI into the world, not one truly transformative - let alone cost-effective - application has been found.
Experts like MIT's Daron Acemoglu and Goldman Sachs' Jim Covello doubt that AI will deliver significant economic benefits. They argue that AI's high costs and current architecture aren't designed to solve complex problems effectively. That it maybe leads to a U.S. productivity growth of 0.5% and a GDP growth of 0.9% over the next ten years. Fairly modest, if not negligible, and certainly not enough to warrant today's investments.
On the other hand, Joseph Briggs, Kash Rangan, and Eric Sheridan from Goldman Sachs believe AI could automate 25% of work tasks, boosting U.S. productivity by 9% and GDP by 6.1% over the next decade. In other words, no one knows for sure!
The report also notes the significant constraints, like chip shortages and potential power shortages, which could hinder AI's growth and scalability due to aging power grids and regulatory challenges. Even so, they still hope for a 'killer app' (but no one knows what that would be). [Link]
To keep things in perspective, a variation on the quote above has been said about the Web in its early stages too. And look where we are today...
Where is the revenue?
For all the funding pouring into AI companies, David Cahn from Sequoia wonders where the revenue is. He describes the vast investments in AI, with a focus on the discrepancies between capital expenditure and actual revenue growth in the AI ecosystem.In his blog, which contains some good charts of estimated and actual revenues, he emphasizes the need for realistic expectations and strategic planning in AI investments. AI will bring economic value, but it will take longer than we think. And forget about the “quick rich” AGI schemes: not happening. [Link]
If you'd rather watch than read, CNBC did a segment on these announcements. They also do a good job of illustrating the story with company financials. The bottom line: the capital expenditure is so large that we can't expect a rate of return. [Link]
Let's get real(istic)
While LLM’s seem magical, their novelty fades quickly, writes Ben Evans. Millions have tried AI tools like ChatGPT, but ongoing usage is limited. He explores the current state of AI technology and the disparity between the high level of interest and actual deployments. Many companies are piloting AI projects, but few have fully integrated these technologies. Evans believes that while AI has transformative potential, the hype often outstrips practical applications. He emphasizes the need for realistic expectations about AI's impact and urges a more measured approach to integrating AI technologies into businesses.[Link]
We vastly overestimate generative AI and we should not assign human capabilities to it, says MIT’s Rodney Brook. The current hype surrounding generative AI, especially large language models (LLMs), leads to unrealistic expectations about their capabilities. He emphasizes the importance of integrating AI into practical applications where robots and humans can collaborate effectively, such as warehouse operations, rather than focusing on creating human-like robots. Brooks also cautions against assuming exponential growth in AI capabilities. [Link]
And what about HR?
And finally, Bain ran a survey and concludes that conversations about generative AI are getting more realistic, moving from hype to genuine assessments. After the initial excitement, they notice a shift towards more realistic assessments, that emphasize practical applications and address organizational readiness. They see five promising use cases: in sales, software development, marketing, customer service, and onboarding. Use cases in legal, operations, and HR (!) appear less successful. [Link]
The chart below from the Bain report shows how expectations about generative AI use have shifted between October 2023 and February 2024. Just look at the gap for HR:
What's next?
Do you think it's time we all take a step back and assess where AI can genuinely help us, rather than assuming it will outperform humans in everything?
AI has been around for ten years and in the HR space, we've been applying it mostly in hiring. Despite some occasional hiccups, that seems to work. Just because generative AI attracts so much attention, it's easy to overlook other forms of AI that is already useful.
I think it's time to balance the hype with practicality, ensuring AI serves to support us rather than replace us. I am so over people who claim AI will do everything, but fail to show us proof. Who say we'll all get Universal Basic Income but don't explain how that will be funded.
We need a clear focus on use cases that provide tangible benefits and improve operational efficiency. AI is not one thing: it consists of many things, from neural networks and machine learning to generative AI. And that means we need to understand how each of these technologies work, and select the right one to add value. Sometimes robotic process automation, not AI, is all you need to complete tasks. Or the use of a scripted chat bot instead of a generative one.
Which means we need to be critical, and not take AI for granted: there's always an alternative. Many pilots fail because the average company is much better off using the AI functionality in vendor solutions, than building something themselves. In short, the generative AI honeymoon is over. Now let's focus on what we really can achieve with new tech.
And PS:
While I was working on this article, I wanted to include a cover image, but after four ugly images failed attempts, Dall-E advised me to hire a professional designer or a local artist:
I found it reassuring that it seems to know its limits and recommended a person would do a better job!😉
Have a great summer, Anita
How to get in touch
My newsletter is free: please think of me when you need an engaging and insightful keynote or workshop!
Or support me by buying my books Equal Pay for Equal Work and How to Select Your Next Payroll. Or get the Dutch translation Gelijk werk? Gelijk loon!
(Remember to switch to your local Amazon store)
If you need HR Tech or payroll advice, you can connect with me here:
What can I do to make the newsletter better for you?
Did you enjoy this newsletter?
Support me and forward it to someone who might like it.