ChatGPT has kicked off a digital arms race, but it’s only the first shot in the huge generative AI disruption that is coming behind it.

Founded in 2004, Facebook took a year to reach a million users. Instagram, which Facebook bought in 2012 for a billion dollars, took two took and a half months.

By contrast, ChatGPT, the breakthrough AI natural language application registered its millionth user only five 5 days after its release on November 30 last year. 

OpenAI – which owns ChatGPT and Dall-E, the popular image-generating tool that provided the images for this newsletter – is already valued at $29bn, which happens to be the market capitalisation of both Dell and HP. If it were public it would already sit comfortably in the middle of the S&P 500.

Microsoft recently revealed it will invest another $10bn in OpenAI (whose backers include LinkedIn founder Reid Hoffman and serial entrepreneur and investor Peter Thiel) over the coming years. According to CNBC, under the terms of its deal Microsoft will be entitled to 75 per cent of profits until its investment is recouped, at which stage it will hold 49 percent ownership of OpenAI.

This is on top of the billion dollars it originally invested in 2019, and an additional $2bn which Forbes claimed Microsoft subsequently tipped into the company over the following years.

Microsoft’s relationship with OpenAI is seen as a potential threat to Google – especially as the tool is integrated into its business applications like Office 365.

Indeed the AI arms race has already begun. Stung by the attention OpenAI – and by extension Microsoft was garnering, Alphabet rushed out the announcement of its competitor, Bard AI. It did not go well. After reports that the social media campaign Google was running to promote the new tool actually demonstrated inaccuracies in Bard, the company’s share price dropped 8 per cent stripping its market capitalisation of $100bn. (That almost made it the most expensive tweet in history.)

To add to Alphabet’s pain, the day after the Bard AI release, Microsoft revealed that it has incorporated OpenAI’s language learning model and capabilities into its Bing search engine.

Meta’s CEO Mark Zuckerberg hasn’t let his recent obsession with adding legs to his avatar in the metaverse blind him to the possibilities either. During an earnings call to discuss its 2022 Q4 results, Zuckerberg told the market: “Generative AI is an extremely exciting new area with so many different applications, and one of my goals for Meta is to build on our research to become a leader in generative AI in addition to our leading work in recommendation AI.”

Why the fuss?

Except perhaps for the AI cognoscenti, generative AI has the same vibe as that bus-sized asteroid that recently whipped past the earth with only a few days’ notice. A bigger bus and a bit more gravity and you would have been facing a cretaceous-level disruption to your annual business plans. It feels like it’s come from nowhere.

In fact, investors have been chasing the generative AI opportunity aggressively now for several years. Last year investor funds spiked from $1.5bn to $2.6bn according to CB Insights, even if the number of deals only ticked up moderately from 105 to 110.

According to the firm, the four largest rounds went to four companies: Anthropic  ($580m Series B), Inflection AI ($225m Series A), Cohere ($125m Series B) and Jasper ($125m Series A).

CB Insights also note that nearly 70% of generative AI companies have so far either raised no external capital, or are only at the seed, angel, or pre-seed stage, and just 16 per cent of companies have achieved Series B or later. OpenAI is the whale in the pond at $29bn, but there are already five unicorns at the generative water’s edge: HuggingFace, Jasper, Glean, and Stability.AI.

So what is generative AI?

Part of a family of tools that falls under the umbrella of Generative AI, ChatGPT has provided artificial intelligence with its broad public “ah-ha moment.”

It is not radically smarter than other AI tools, but it is much easier to use and understand and as a language tool its utility is ubiquitous, unlike many other tools.

AI bubbled away in the lab for decades and even when it started filtering into wider business use it was held back by the need to find expensive and scarce data scientists to manually build the parameters for the experiments the machine needed to run in order to “learn”.

There was a huge leap forward in late November 2017 (coincidentally, almost five years to the day before the public release of ChatGPT) when Amazon launched Sagemaker.

Why was Sagemaker a big deal? Let’s ask ChatGPT

“Amazon SageMaker is a cloud-based machine learning platform that makes it easy for developers and data scientists to build, train, and deploy machine learning models. In simpler terms, it's a tool that helps people use artificial intelligence to make predictions based on data, without needing to have advanced programming skills or access to expensive computing resources.”

Sagemaker automated the most time-consuming, and least valuable aspect of data science and released a wave of AI fuelled innovation in business. 

But it was still niche, and rarely extended beyond the walls of IT or marketing departments. It was often little understood, poorly deployed and often hard to scale beyond proof of concept. Executives often wanted AI to re-enforce their personal biases, rather than to usher in an era of data-led decision-making.

For consumers, the utility of AI was even more remote. If consumers ever personally encountered it, it was probably via a chatbot whose principal purpose was to automate repetitive service routines and save companies money, rather than to improve customer service.

Generative AI, however, is new and genuinely transformative. But what is it? 

Let’s ask ChatGPT.

“Generative AI refers to artificial intelligence models that are designed to generate new content, such as images, text, audio, or video. It works by using existing data to learn patterns and then using that knowledge to create something new that resembles the original data. Think of it like a computer program that can "imagine" new things based on what it has seen before. For example, a generative AI model trained on pictures of faces could generate a new, never-before-seen face that looks similar to others it has seen.”

Actually, it’s smarter than that. It generates content based on the parameter the user defines. 

When we asked ChatGPT to describe generative AI, we instructed it to do so in language an ordinary person would understand. But what if we wanted to tailor the answer to the investment community?

“Generative AI is a subfield of artificial intelligence that focuses on the development of models capable of creating new and original content. These models use machine learning algorithms to analyse vast amounts of data and then generate outputs that are similar in nature to the input data. The outputs produced by generative AI can be in the form of images, text, audio, or video.

Generative AI has numerous applications across various industries, such as the creative arts, advertising, and media. The technology has the potential to significantly impact traditional content creation processes and is attracting significant interest from investors as a result.

Investors should be aware that generative AI is still an emerging technology, and there is a significant amount of research and development required to bring these models to market. However, the potential for disruption and the creation of new opportunities in various industries make generative AI a promising area of investment.”

Commercial application

There are already an array of generative AI applications in the market, of which ChatGPT is simply the most famous. 

Areas where it is being applied include:

• Generative Adversarial Networks (GANs) for image synthesis and manipulation

• Text-to-Speech (TTS) systems for voice synthesis

• Music synthesizers that generate original music

• Text generators that can produce articles, stories, or even poetry

• Video generators that can create new video content based on existing footage

What’s it not?

While ChatGPT is seen as a threat to Alphabet’s search business (especially with Microsoft behind it), it is not a traditional search engine like Google or Bing. You can ingest words and data as text into your queries however, and OpenAI is working to publish an API that companies can write to, which will make this easier.

But as it works today, OpenAI stopped ingesting information into the data model underpinning ChatGPT in 2021, so it's next to useless for grabbing the latest English Premier League scores, or Meghan Trainor TikTok meme. 

It does, however, offer the possibility that it could disrupt search markets by redefining them.

We asked ChatGPT to tell us how it’s different from Google. Its initial answer contained just a hint of hubris, extolling its own virtues while damning Google with faint praise. That gave us an idea. We decided to amp up the generative snark and asked it to answer in a teenage voice:

"Think of ChatGPT as a smart friend who can hold a conversation with you and answer your questions, while Google Search is like a library with lots of information, but you have to know exactly what you're looking for.

ChatGPT can understand what you're saying and respond to you in a way that's easy to understand, just like you would talk to your BFF. Google Search is good for finding specific information, but you have to use the right words to get what you need.

So, if you're looking for a chatty pal who can talk about anything and everything, go for ChatGPT. But if you need facts and information fast, Google Search is the way to go!”

Uhuh. There you have it. Teen-voice ChatGPT says ChatGPT is sick, but Google is totes chuegy. 

Google may be the OG but ChatGPT is the Top G, as the kids would say. 

Note: image courtesy of Dall-E 2: “a painting in the style of Norman Rockwell on the benefits that generative AI can deliver to society.”