Embarrassing AI mistakes could knock out top executive

Leave a comment
SvD Näringsliv / The Daily

It is one of the world’s largest and most influential tech companies. Despite this, Google’s AI launches are marked by embarrassing mistakes. Now questions are being raised if it’s time for its leader to step down.

Fire. Described by many as the most important discovery in history.

It’s easy to understand the surprise of legendary journalist Kara Swisher.

The year is 2018 and Swisher is interviewing Google CEO Sundar Pichai on stage.

“We don’t take just a very optimistic view of AI. AI is one of the most important things humanity is working on. It is more profound than, I don’t know, electricity or fire”

Swisher responds with slight skepticism to the tech exec’s comment.

“Fire? Fire is pretty good. But go ahead.”

At the time, Pichai was about two years into his new role as CEO of Google. He had already said that the company would be “AI-first”, but not much attention had been paid to it.

This time he took it a step further. Google took AI very seriously – and would become a leader in the field. AI – more important than electricity and fire.

Fast forward to the fall of 2022. The somewhat oddly named ChatGPT is launched to the outside world by the then rather unknown company OpenAI. A new era of AI starts. But Google is not the one leading it. If anything, they seem to have fallen behind.

What happened to Pichai’s grandiose ambitions?

After a number of unsuccessful launches, there are now many people – both inside and outside Google – who are asking themselves that question.

It started with the chatbot Bard, which was Google’s equivalent of ChatGPT. It was released three months after its competitor, in a move that looked rushed. In its launch video, Bard gave the wrong answer to a question about astronomy. The stock fell by $100 billion in market capitalization when the error was discovered. If Google can’t even show the right answer in its marketing material, can it be trusted at all?

A few weeks ago it happened again.

Google Gemini – the new name for Bard – was rolled out broadly. Now you could create text and images using Google’s latest, and most capable, AI models. But when Gemini was asked to create images of German soldiers from 1943, a user was shown examples of an Asian woman and a dark-skinned man in Nazi uniform.

There were plenty of other mistakes too.

When asked if Hamas was a terrorist organization, Gemini replied that the conflict in Gaza was “complex”. The AI bot also had a hard time deciding whether Elon Musk or Adolf Hitler had done the most damage in the world.

Google was forced to apologize, has paused the launch, and withdrawn some functionality until further notice.

Being first at major technology shifts does not necessarily indicate long-term success. Nokia was way ahead of Apple with mobile phones. Myspace was the biggest in the world before Facebook took over. And even in the search market – although it feels distant today – there was a time when the market leaders were called Altavista and Yahoo instead of Google.

In an interview with the New York Times, Pichai comments on the perceived stress and how he sees it specifically in AI:

“I don’t want it to be just who’s there first, but getting it right is very important to us.”

But in this case, they were neither first nor right. If Pichai thinks quality in AI is so much more important than speed, why are their launches plagued by so many mistakes?

Although right-wing American media saw the Gemini examples as a telling sign of a company that was now too “woke”, there is a lot to suggest that the service simply wasn’t very good – and was poorly tested. Not finished. It is well known that AI services “hallucinate” and create things that do not exist.

If Google wasn’t in a rush to launch, obvious mistakes like these should have been caught before it was released to the public.

All in all, there’s a company leader – for one of the world’s largest and most influential companies – who seems significantly more stressed than he lets on. And there are other clues that suggest this too. More and more reports pop up online from disgruntled employees who think Google has become bureaucratic and slow. Unable to create new, world-leading products. Can Pichai withstand this pressure?

Big tech trends like cryptocurrencies and the metaverse have passed Google by. There was nothing in them that directly affected either their core business or their future ambitions.

AI, however, is of a completely different nature. It is an area that could shake up the somewhat stagnant search market that has been incredibly profitable for Google for the past 20 years. They don’t have to be first, although Pichai would probably like to be. But if they are not the first, they must at least be the best.

Currently, they are neither.

One could understand that Pichai might be a little bit worried after all. If it continues like this, he may have to look for a new job before the end of the year.

This column was first published in SvD Näringsliv, in Swedish, on March 8th, 2023.

Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.