A first of its kind halloween edition on my blog and I thought I theme it a bit that way.
Now, talking about AI there are many things that seem scary to some and others. Mostly such feelings are around how it might impact employment or changes in the workplace, but also how it might impact our public discourse. I am not necessarily disagreeing with some of the fears – as well as the opportunities. Today though I don’t want to write about this, but a different scary AI topic.
ChatGPT – almost one year in
We are arriving soon at the 12 months mark when ChatGPT was released to the public. Last November 30th it was brought to public release and started a hype that since has not really stopped. It was the “App” with the fastest growing user numbers ever with reaching 1 Million users after 5 days and 100 million active users by January 2023. No other App ever managed that. But that wasn’t it. ChatGPT with the help of Microsoft made it quickly into the enterprise market and many of us are experimenting with it or other Gen AI solutions since. Also in the enterprise market this AI revolution has taken offices by storm – may it be IT professionals or functional professionals across all functions – Sales, Marketing, Finance, others – and of course Human Resources. The ideas and applications are universal and exponential. This is also what makes AI so different from the Metaverse or Blockchain. As I already wrote here, Metaverse and Blockchain were interesting new ideas and technologies…in search of a business problem to solve. And they are still in this same position years after they saw the light of day. (Gen) AI came with a set of addressable business problems and has identified many more to solve.
The potential of (Gen) AI
I am sure many of you have experimented with (Gen) AI in the last 12 months as I have and have seen the potential it has and the immediate applicability. We have also seen the flaws and over-hypes. But at its core it was and is a revolutionary technology that transforms the way we do business, the way we do HR. May it be in the Experience space where Gen AI significantly improves the way regular users can interact with HR technologies, even if they only need to use it once or twice a year. AI has solved this issue by “taking users by the hand”. May it be in the skills space where it not only can identify skills you most likely have based on your CV and previous positions, but also has a superior matching quality to find the right job, the right opportunity for you – or helps you with understanding what skills and experiences you need to gain before you have a chance of getting your dream-job.
Sometimes it felt like (Gen) AI is the solution to almost all issues we have in today’s HR world. Of course that is not true and of course as much as I did, many of you saw some of our test & learn concepts fail or underdeliver against our expectations. But we carried on to identify new ones – especially motivated and cheered by all the Tech companies that had AI technology readily available. And who had not started to get ready. It was a hectic gold-rush for solutions and ideas in the last 12 months – and slowly but surely we wake up to a scary AI reality. A reality that quite frankly we should have seen coming – and for sure some of us saw coming. Question though is if you expected that level of scaring potential.
Scary AI
What am I referring to? – while most of the vendors and partners did let you experiment and even pushed for new ideas for experimentation heavily these last 12 months at either no costs or very low costs, the boomerang now comes back. AI is expensive – especially Gen AI. Each training run for a Large-Language Model (LLM) can cost between $2-$12 Million. It took OpenAI more than $1.4 Billion to get to ChatGPT public release. And the actual usage costs of these models once ready are not cheap either. There is a good reason why NVIDIA’s share price performed +205% over the last year. NVIDIA is the main supplier of the GPU chips necessary to train and run LLMs. But not only NVIDIAs share price performed well – look at Microsoft with an increase of 45%, closing to Apple as being the most valuable company in the world. And of course these investments and expectations are asking for “re-financing” or in other terms, the costs of using (Gen) AI models in regular production applications will cost a scary amount of $$$ in order for these companies to get their investment back and to confirm their promises to the stock market. And guess who should pay that price? – yes, you and me🙂
(Gen) AI is here to stay – there is no doubt about it and there will be some adoption happening. The question is how much and how fast will it grow. We are still in our hype-cycle with (Gen) AI and I believe the costs of (Gen) AI will send us into the trough of disillusionment. How long we stay there depends on two things really
(1) how high the price will be that the (Gen) AI vendors will settle for and if they are willing to defer their returns a bit longer – and yeah, I get it, even the high prices have often not too high margins as costs need to get under control first and
(2) if the price to pay is at least equal, but better lower compared to the additional value or cost efficiency that companies are able to realize.
We are on the down-turn of the hype-cycle, but how long we will stay there and how deep we will go is an interesting question. I have no doubt that we will reach a rather high plateau of productivity – but when? What do you think?
Chief HR Officer 