How (not) to talk about AI with your shareholders

When robots break down your front door, how will you come?

Just a few months after ChatGPT arrived on the scene with the promise of making moderately bad writing more accessible, many businesses are coming to terms with their apparent nemesis.

While a couple of nodes:

— resistance to learning models that underpin technology
– the sheer nastiness of the inputs/outputs

are resolved, the capital markets are not wasting time tearing off the clothes. According to the FTlast week:

Stocks in the education sector have fallen sharply [on May 2nd] while investors are betting that AI could upend business models following a revenue advisory at edtech company Chegg.

California-based Chegg, which provides online study guides, admitted on Monday night that a “significant spike in student interest” in the AI ​​chatbot ChatGPT was starting to hurt its sales.

Chief Dan Rosensweig said:

So this is not something that falls out of the sky. It’s just an acknowledgment that there’s been a technological shift, and we need to prepare and adapt our business and aggressively pursue that and adjust our cost structure to do so. And we are doing all of this now.

Chegg’s share price was more or less halved by the end of the day – its problems are all the more dire because the company has been quite quick to embrace AI (more info from our FT principal colleague Richard Waters Here).

Other companies are now trying to avoid a similar scenario. Yesterday morning, the Pearson education group held a call with analysts and the media specifically to discuss AI (stocks closed up about 1.75%), while RELX, which runs several news activities, is hosting one this afternoon.

ChatGPT’s success has sparked a lot of complaints/”we actually did it boring first” from big tech companies, and Robin wrote about research quantifying the impact of generative AI hype here. Tl;dr, that’s pretty chunky.

But right now the tech giants aren’t the big story. It’s the (relatively) welterweight ranges where AI can be really disruptive, and cautionary tales like Chegg’s can push smaller players to be more proactive in communicating the risks of AI.

Citi’s European Internet and Media team has been looking at this communications challenge, as part of an effort by analysts to map AI risk within the stable of stocks they cover. They write:

Our conclusion is that while Generative AI holds promise to impact a decent percentage of the companies under our cover, it likely brings with it as many opportunities as risks. We are especially focused on those companies with domain-specific/proprietary skills/datasets and institutional client relationships, and take a contrarian position on AI risk/reward.

Using a four input system:

1. First-order implications: direct impact of AI in the short term.

2. Second-order implications: long-term direct impacts of AI.

3. Cost Implications: How AI Will Affect Business Costs.

4. Industry demise risk: A separate assessment of long-term impacts that includes exposure to other industries that may be disrupted.

Here’s what they came up with:

Citi says:

In short, this is somewhat arbitrary, but in short we select an AI risk score of 1.0 as the point at which it should be considered by investors.

Aggregated and plotted, it looks like:

Expanding the analysis company by company, Chegg reigns supremely fucked with a score of 2.5:

Of course for Chegg/whoever holds his stock, this is absolutely pointless because his price has already collapsed. But is there a lesson to be learned about risk communication? Guy.

City:

For those companies that are obviously in the “least affected” bracket, we believe there is relatively little that can be done to address broader market-wide concerns about the potential disruption of technologies based on generative AI.

For those companies potentially most affected, however, something is awkward. In a nutshell it is this: is it talked about or not?

Analysts have been crunching mentions of AI-related keyword buzzwords since Nov. 1 and the stock price’s performance since March 14, in an effort to make a connection:

In that respect, Chegg looks a bit like the awkward-looking dude in a zombie apocalypse movie who keeps wondering how much you need to be bitten to get infected.

Analysts write:

While the true correlation (and causation) is likely to be between actual “AI perceived risk” and stock price performance, to the extent AI mentions it is a “perceived risk” signal, it also clearly shows some relationship . What is perhaps striking, however, is that talking about it much does not seem to have helped Chegg massively. That said, perhaps it also demonstrates that being proactive to some extent can be beneficial.

It strikes us that this is using a lot of words to say “we have no idea.” And that’s before we even get to the weirdness of the timing: Sure, ChatGPT-4 launched in mid-March, but the incessant “AI is coming for you” talk has been going on since (at least) the end of last year.

Undoubtedly, the great revolving minds of the flack financial world will be hard at work developing ways to keep their clients from cheating themselves. While we wait, here’s what ChatGPT-4 recommends:

Actually don’t worry about the spinners, you have this…

Further reading
— Fourteen Reasons Why Pop AI Won’t Eat Itself