IBM Pausing new hiring
It's not the base, it's the change
Before I start this week’s essay, a correction: On Friday I shared how I had used Bing’s AI search to summarize for myself the results of Meta’s quarterly earnings call. The first “fact” the AI gave me was “Meta Platforms reported a strong quarter with revenue of $38.6 billion, up 31% year-over-year, and net income of $11.2 billion, up 17% year-over-year12” — with a footnote no less! And yet the actual quarterly revenue and net income were $28.6MM and $5.7MM (up 3% and down 24% respectively. Summarized nicely here). It is well known that generative language models “hallucinate” and get facts wrong, but it happens less frequently for GPT-4, and it is even more rare when you ask a model connected to the internet for references (which was what happened in this case). It doesn’t mean you should not use the models, but it does mean you should do a better job than I did verifying their references. Onto today’s post:
On May 1st, IBM announced a (partial) hiring freeze. From Bloomberg:
International Business Machines Corp. Chief Executive Officer Arvind Krishna said the company expects to pause hiring for roles it thinks could be replaced with artificial intelligence in the coming years.
Hiring in back-office functions — such as human resources — will be suspended or slowed, Krishna said in an interview. These non-customer-facing roles amount to roughly 26,000 workers, Krishna said. “I could easily see 30% of that getting replaced by AI and automation over a five-year period.”
That would mean roughly 7,800 jobs lost. Part of any reduction would include not replacing roles vacated by attrition, an IBM spokesperson said.
When I was at McKinsey around 2006, I did a lot of work in lean operations. We would go into retailers, observe and track everything employees were doing, and then use a playbook of to change the operating procedures or training to improve efficiency. The result was almost always significant cost savings from fewer employee hours, combined with improved customer service. There was rarely a need to layoff employees, instead the current employees were repurposed and attrition took care of the rest. But at the end of the year total payroll costs were significantly lower.
In some ways “lean operations” could be traced back to Henry Ford and the first assembly lines, but modern lean is generally attributed to Toyota’s management system from the 1980s and then popularized in America by James Womack’s book, “The Machine That Changed the World”. Whatever the origin, it definitely had been around a long time by the time I was using the techniques to fix shelving practices in 2006. It has historically taken a long time for new techniques and processes to work their way through American business until they become “table stakes”.
Generative AI is a few years old now, but the current capabilities are less than two months old (depending on what you define as those capabilities — more stuff is happening every day). It is not surprising that most companies still aren’t using it. It will take multi-million dollar McKinsey engagements to get many companies to fully imbrase the implications and cost savings opportunities brought by AI.
Contractors are the first at risk, but I don’t see anyone losing their job… yet.
But what will happen is that future head count will be impacted. If you had a “content creator” on your hiring plan, it is unlikely to be approved. But unless your company is doing a layoff already, you are likely not going to be asked to fire your current content writer.
The same logic applies for companies that are going to be disrupted. Clegg is a “homework helping tool” that has a database of answered to the questions schools have asked in the past. Why pay for Clegg, when GPT-3.5 can give you everything you need for free?
From the earnings call on Monday:
In the first part of the year, we saw no noticeable impact from ChatGPT on our new account growth and we were meeting expectations on new sign-ups. However, since March we saw a significant spike in student interest in ChatGPT. We now believe it’s having an impact on our new customer growth rate.
Fortunately, we continue to see very strong retention rates, suggesting that those students who already understand the value of Chegg continue to choose us and retain us at high rates. We are also expecting a positive recovery in enrollment trends, which historically would be good news for Chegg. Because it’s too early to tell how this will play out, we believe that it’s prudent to be more cautious with our forward outlook. Therefore, we intend to provide only the next quarter’s guidance at this time and Andy will walk you through those details shortly.
In Q1 ChatGPT had no impact. By March and April it started hurting new sign-ups, but they “continue to see very strong retention rates”. Users that are comfortable with Clegg are not “firing” the company, but non-users who are looking for a solution are comparing Clegg to ChatGPT and are deciding the free option is good enough.
Over time a lot of jobs and companies are in trouble, but the impact will first be seen on growth rates, not base rates. That’s the metric to watch for.
Keep it simple,
Edward