The AI Arms Race intensifies the bizarre dynamics
Fourth in a series of articles that consider why this particular moment in the tech industry is so strange.
This is the fourth in a series of articles about the intense strangeness of our moment.
In the previous articles in this series, we considered some things that make the advent of consumer-facing artificial intelligence (AI) apps seem so weird: the accelerating pace of change; the declining costs of parallel processing as an accelerant to the deployment of large language models (LLMs); the unreliable nature of conversational AI and the split personalities that occasionally surface during consumer interaction; and the unprecedented experience of using a consumer-facing product that learns, improves and occasionally demonstrates a new capability that no human programmed.
We’re breaking these into separate topics because we’d like to dispel some of the fear and hype fog that make the launch of consumer-facing artificial intelligence so confusing.
The goal is to peer past the hype to gain a better understanding of what’s happening. You can find the other articles in this series on the Substack archive for The Owner’s Guide to the Future.
Now let’s consider the competitive business dynamics that have intensified the pace of innovation in AI.
Weird Thing #7:
Instead of proceeding with caution, tech firms are engaged in escalation.
OpenAI does not operate in a vacuum. Competitors are racing to catch up. To stay ahead of the pack, OpenAI will make GPT-4 available via API to a growing number of companies. This will accelerate innovation, and it could turn OpenAI into the next big web platform. It also increases risk.
According to MIT Technology Review:
“GPT-4 may be the best multimodal large language model yet built. But it is not in a league of its own, as GPT-3 was when it first appeared in 2020. A lot has happened in the last three years. Today GPT-4 sits alongside other multimodal models, including Flamingo from DeepMind. And Hugging Face is working on an open-source multimodal model that will be free for others to use and adapt.
Many other companies are waiting in line: “The costs to bootstrap a model of this scale is out of reach for most companies, but the approach taken by OpenAI has made large language models very accessible to startups,” says Sheila Gulati, cofounder of the investment firm Tola Capital. “This will catalyze tremendous innovation on top of GPT-4.”
(OpenAI co-founder Ilya) Sutskever suggests that going slower with releases might sometimes be preferable: “It would be highly desirable to end up in a world where companies come up with some kind of process that allows for slower releases of models with these completely unprecedented capabilities.”
But the biggest battle is not the one between OpenAI and other startup AI firms.
It’s the conflict happening between Microsoft and Alphabet.
The AI arms race may destabilize the tech balance of power
Microsoft CEO Satya Nadella single-handedly elevated AI weirdness to a new level in March. He called Google out to a showdown. Or maybe it was a dance competition? Read on.
In an interview with The Verge about the new AI-enhanced version of the Bing search engine, Nadella said, referring to Google:
“They’re the 800-pound gorilla in this… and I hope that, with our innovation, they will definitely want to come out and show that they can dance. I want people to know that we made them dance, and I think that’ll be a great day."
Let’s break this down. First, the image of an 800-pound gorilla dancing is strange. And when you watch the video of this interview at the link above, you’ll also notice how peculiarly intense Nadella seems to be when he discusses Google. At 14:00 in the video, it seems like he is doing his best impression of an Austin Powers supervillain. He smiles mirthlessly after boasting about making Google “dance”.
This is not normal. The CEOs of giant public companies are typically bland and boring — by design. Their communications teams advise them to refrain from calling out competitors by name during an interview.
But Nadella went out of his way to damn the CEO of Alphabet with faint praise.
Who does Nadella think he is? Does every tech CEO have Musk envy?
Nadella knows that Microsoft is starting from behind in this race. Microsoft’s Bing search engine attracts roughly 10% the traffic that Google routinely gets. In artificial intelligence and robotics, Google is also ahead of Microsoft, having acquired DeepMind and launched the first large scale AI apps more than a decade ago. Google researchers pioneered the original transformer model that underpins ChatGPT.
But Nadella also remembers that Microsoft once had a thriving franchise in the Windows operating system and the Office PC software suite. At least, they did until Google’s scorched-earth tactics with Android, Chrome and Google Docs demolished the business model for those venerable businesses.
Now Microsoft is poised for revenge. Some reports claim that Nadella is so driven to beat Google in AI that he is cutting corners on ethics and safety.
Nadella’s timing is excellent. Google is hamstrung in at least three ways.
First, longstanding internal debates among employees oblige to Google to proceed with caution. Google’s AI ethics researchers were among the first to call attention to the weaknesses and potential problems in large language models. Google famously fired the AI ethics team that reported this issue. After that, more than 2600 Google employees and 4300 industry supporters signed a letter of protest. Then two more AI ethics researchers quit in a sensational manner. If Google now launches an unproven product that fails catastrophically, or causes societal harm, they can’t say they weren’t warned. The management has silenced or ousted those who sounded the alarm.
Second, Google is under the scrutiny of the US FTC and DOJ as well as European regulators for a range of issues, from violations of privacy laws to a federal antitrust lawsuit. This is not a favorable time to leverage an existing monopoly in order to dominate a new and emerging category of computing.
Third, Google can’t take risks with the core search product the same way that Microsoft can. Few people rely upon Bing search: Microsoft could take a wrecking ball to the home page and there would be no complaints. Most people wouldn’t even notice.
But Google’s home page attracts 8.5 billion visits a day. On average, Google handles 99,000 searches every second. Changing jet engines in mid-flight is suicidally risky under any circumstances, but in Google’s case the entire world (outside of, say, China and North Korea) depends upon it.
Google faces the classic Innovator’s Dilemma, coupled with a DOJ antitrust suit.
Not that they haven’t tried to counter Microsoft. Google made a lame effort in February, and it was a spectacular flop. It backfired so badly, the search giant must now proceed with utmost caution.
Baited by Saytella’s taunts, and against their better judgement, on February 8, Google ran an ad on Twitter that provided the public with a sneak peek at Bard, the search giant’s response to ChatGPT + Bing. But one of the answers generated by Google Bard was factually incorrect.
It was a rookie blunder that could have easily been fact-checked by, well, …searching on Google. The stock slid 9% and $100 billion was erased from Alphabet’s market cap in a single day of trading. Microsoft shares rose 3% on the news.
Google has since attempted to recover, launching tools for 3rd party developers to integrate Bard into their apps and making vague noises about features that may or may not be released in the future.
But the damage has been done.
The widespread perception is that Google has missed the first round of the chatbot wars, and momentum is slipping away. On April 16, the New York Times reported “panic” inside Google when Samsung threatened to switch to Bing as the default search option.
Alphabet is struggling to maintain AI safety and transparency while racing to deploy. This month a new transparency page called “Bard What’s Up” launched, to keep the public informed about the company’s AI deployment.
The pressure on Alphabet CEO Sundar Pinchai is undeniable. He faces rumors that he will be pushed out of the CEO role. If Bard continues to stumble, those rumors might turn into shareholder votes.
That’s why you can expect the next round of competition between Microsoft and Google to become even more intense. The stakes for these two giant companies may be as high as life and death. Between the CEOs, it’s personal.
OpenAI, Microsoft and Alphabet are not the only ones in the fight. During the first three months of 2023, more than 150 AI apps were launched.
China’s Baidu has launched its own AI chatbot, known as Ernie. Baidu is considered the leader in the race in China to develop the first scalable rival to ChatGPT.
On April 13, Amazon entered the fray with a new suite of tools that will help companies create and manage their own chatbots and image-generating systems.
Amazon Web Services introduced a new cloud offering known as Bedrock which will provide customers access to foundation models that will enable companies to develop their own AI systems and apps built upon their own data. Bedrock provides customers with the ability to customize large language models, including Amazon’s proprietary LLM foundation models, known as “Titan”, as well as models from third-party providers including Anthropic, Stability AI, and AI21 Labs. This package will enable Amazon AWS customers to test AI technologies without the need to invest in massive data center infrastructure.
The arms race is on.
Key Insight: Competitive “winner take all” dynamics and an ancient blood feud between Microsoft and Google have contributed to the AI acceleration. Nothing will constrain the leading companies: they may be tempted to sidestep ethical guardrails, safety measures and all other precautions in their pursuit of market share and integrations with the broadest number of developers and end customers.
The entry of Amazon, Baidu as well as 150 startup ventures means that there will be plenty of competition in this sector. This competition will contribute to the accelerated deployment of AI.
This is the fourth in a series of articles that examine the deep weirdness of this particular moment in the technology industry. We’re considering various aspects of the deep weirdness of consumer-facing artificial intelligence. The goal is to peer past the hype and the fear mongering to gain some insight into what is really happening and why it matters. If you’ve enjoyed this article, why not share it with a friend?