With guest writer ChatGPT

OpenAI LLC and ChatGPT have taken the world by storm. Microsoft Corp. reportedly is investing an additional $10 billion in the startup.

But in our view, though the hype around ChatGPT is justified, we don’t believe OpenAI will lock up the market with its first-mover advantage. Rather, we believe that success in this market will be directly proportional to the quality and quantity of data that a technology company has at its disposal, and the compute power it has to run the system. This market is unlikely to display winner-take-all dynamics and will probably evolve in a more fragmented fashion than cloud computing.

In this Breaking Analysis, we unpack the excitement around ChatGPT and debate the premise that the company’s early entry into the business may not confer “game over” advantage to OpenAI. To do so, we welcome CUBE collaborator Sarbjeet Johal and John Furrier, co-host of theCUBE, SiliconANGLE Media’s video studio.

What is ChatGPT in 50 words or less

Rather than asking our guests, we went right to the source and asked ChatGPT itself.

ChatGPT: Please explain ChatGPT in fifty words or less:

ChatGPT is a state-of-the-art language model developed by OpenAI that can generate human-like text. It can be fine-tuned for a variety of language tasks such as conversation, summarization, and language translation.

We asked our guests to add color to ChatGPT’s answer. We then fed the transcript of the guests’ answers into ChatGPT. Here’s how ChatGPT summarized the conversation:

John Furrier and Sarbjeet Johal discuss ChatGPT as a large language model, which utilizes the transformer mechanism to focus on the given prompt and the previous answer. They mention that it can be used for writing, marketing, copy, brainstorming, guest list, and more. However, they also note that it is not always accurate and there is room for improvement. They also mention that Microsoft has invested in it and there may be future applications in word processing or operating systems. They also mention examples of people using it for interviews, song lyrics and in-depth philosophical conversations.

Listen to the actual conversation and compare how well ChatGPT captured its essence.

Unlimited possibilities for AI and natural language processing

For those who haven’t played with ChatGPT, it’s pretty remarkable. One of the best examples of mind-blowing use cases was Sam Charrington from the This Week in ML podcast. We found this thanks to Brian Gracely on one of his Cloudcast podcasts. Ben did a segment called Will ChatGPT Take my Job? with ChatGPT.

What Sam did was prompt ChatGPT to interview ChatGPT and he simply gave the system the prompts. ChatGPT automatically generated the questions, which Sam then fed into ChatGPT to generate the answers. Then he ran the questions and answers into this avatar builder shown above, sped it up two times so it didn’t sound like a machine, and voila. Amazing.

We asked John if ChatGPT was going to take over our jobs as CUBE hosts and we asked for Sarbjeet’s opinion. We fed their answers to ChatGPT and here’s what it returned:

John and Sarbjeet discussed an example of using ChatGPT in a podcast where it was prompted to interview itself, with the responses being fed into an avatar builder and sped up to sound like a conversation between two people. John sees value in using ChatGPT for tasks such as writing marketing copy, brainstorming guest lists, and providing answers to specific questions. However, he also acknowledges that the answers generated by the model are not always 100% accurate and require some editing. Sarbjeet also mentioned an example of listening to a fireside chat with ChatGPT, where it was able to give detailed answers in a conversational format. Both agree that while current AI models like ChatGPT are democratizing intelligence, they are not yet able to create new intelligence or provide a competitive advantage in certain areas.

Listen to the actual conversation and compare it to ChatGPT’s answer above.

Is ChatGPT a Netscape moment (look it up, kids)?

What about this notion that Futurist Roy Amara put forth – Amara’s Law — that we’re showing below.

We asked John and Sarbjeet if that’s the case with ChatGPT. Here’s how ChatGPT summarized their answers:

John Furrier and Sarbjeet Johal discussed the idea of Amara’s law, which states that we tend to overestimate the impact of technology in the short term and underestimate it in the long term. Sarbjeet thinks that this is true for ChatGPT and that people become less impressed as they use it more and start to see its limitations. John, on the other hand, believes that ChatGPT is a unique and powerful technology that has been underestimated in the short term and will exceed expectations in the long term. He compares it to the early days of the internet, where it was initially dismissed as a toy for kids but has since had a huge impact on society. He notes that some experts are pooh-poohing ChatGPT in the short term, but others see it as a game-changer. They both agreed that the hype is bifurcated.

Listen to the actual conversation on near-term and long-term hype and compare it to ChatGPT’s answer.

Will OpenAI get first-mover advantage from ChatGPT?

Let’s get to the heart of the premise and possibly the debate for today’s episode. Will OpenAI’s early entry into the market confer sustainable competitive advantage to the company? History says maybe, but the consensus of the panel is: Probably not.

If you look at the history of the technology industry, it’s littered with first-mover failures. Altair, IBM, Tandy, Commodore and Apple all took a back seat to Dell, which came on the scene years later with a better business model.

Netscape was all the rage in silicon valley with the first browser.

AltaVista was the first search engine to index full text.

Friendster and MySpace came before Facebook.

The iPhone certainly wasn’t the first mobile device.

There are many failed examples, but there are some recent successes, such as Amazon Web Services in cloud, and Twitter you could argue had first-mover advantage, and there’s bitcoin in cryptocurrency.

We asked the panelists: Will OpenAI make it to the list on the right with ChatGPT?

Here’s how ChatGPT summarized their responses:

John Furrier and Sarbjeet Johal discussed whether OpenAI’s early entry into the market with ChatGPT will confer a sustainable competitive advantage for the company. They acknowledged that the technology industry is littered with first-mover failures, such as Altair, IBM, Tandy, and Commodore in the PC game, and Friendster and MySpace in the social media space. However, they also noted some recent successes like AWS and cloud, Twitter, bitcoin and crypto, and smartphones. John thinks that OpenAI as a company probably won’t make the list of successful first-movers, but the category of large language model AI that they are working on will be on the right-hand side. Sarbjeet agreed that OpenAI’s early entry into the market with ChatGPT will not confer a competitive advantage to the company, as Google has been working on similar technology and other companies will likely compete as well. He also pointed out that ChatGPT is still vulnerable to mistakes and has a lot of questions around the source of its content, and therefore it is a risky move by OpenAI.

Listen to John Furrier and Sarbjeet Johal debate whether OpenAI will sustain first-mover advantage or go the way of Friendster and MySpace.

The hyperscalers dominate AI. Will they challenge ChatGPT?

Let’s take a look at some of the data from Enterprise Technology Research and you’ll see the dominance of the hyperscalers in enterprise AI.

The slide above shows Net Score or spending momentum on the Y axis and Pervasion or market presence in the ETR survey on the X axis. The table insert informs as to how the dots are plotted.

The most relevant key takeaway is the hyperscalers are dominant. And though there are some others on the list, you see above the big three are all over this space.

Yes, there are specialists such as C3.ai, DataRobot, SparkCognition, H2o.ai, Anaconda, Dataiku, with Oracle and IBM in the game. But their mindshare among enterprise buyers pales in comparison with the big three cloud players.

So with that as a backdrop…

Who wins the conversational AI battle?

You’ve got the big cloud and internet giants all poised to go after this opportunity. Microsoft investing up to $10 billion, and Google has evidently sounded the code red. There’s Apple, of course, and several alternatives on the market today such as Chinchilla, Bloom or Jasper. Meanwhile, Federal Trade Commission Chair Lina Kahn and the governments around the world loom large – the U.S., the EU, China – all taking notice before the market has really coalesced around a single player.

Here’s how ChatGPT summarized the conversation on this topic:

The guests discussed the competition for AI and natural language processing, noting that big tech companies like Microsoft and Google are investing heavily in the field. They also mentioned smaller companies like Chinchilla, Bloom and Jasper as potential competitors. They discussed the potential involvement of the government in regulating the field, with one guest expressing concern that the government may label big tech as “evil” and that their actions are politically motivated. Another guest pointed out that government intervention in the past has failed to stop market forces from determining success. They also talked about Microsoft’s investment and how it will be applied, with one guest suggesting it will be used in their productivity and developer tools. They also point out that the speed of technological change is now much faster than it used to be and that two years is now the equivalent of 10 years.

Listen to John Furrier and Sarbjeet Johal riff on big-tech competition, governments the pace of innovation.

How will ChatGPT make money?

What’s the business model for ChatGPT? Here’s how ChatGPT summarized the conversation between John and Sarbjeet:

The VCs all have FOMO – $100M gets you in the game!

Image: RATOCA

At our Supercloud2 event this week, Howie Xu, a leading AI expert, was asked how much money it would take to clone ChatGPT. Here’s what he said:

One of the VCs actually asked me the other day, “Hey, how much money do I need to spend invest to get a, you know, another version of OpenAI sort of of the level?” I said $100 million is the order of magnitude that I came up with. You know, not a billion… not $10 million… around $100 million.

Listen to Howie Xu’s estimate as to what it would take to clone OpenAI and ChatGPT.

Here’s ChatGPT’s summary of the conversation between John and Sarbjeet:

In this conversation, the speakers discuss Howie Xu’s claim that it would take approximately $100 million to clone ChatGPT. They express surprise at the figure, with one speaker suggesting it is an “astoundingly low figure.” They also mention that the barriers to entry in terms of cost are lower than that of competing with AWS. They also mention that it may take a lot more than a hundred million to run the inference machine and to maintain the business, and also the importance of a flywheel effect for businesses to maintain success and growth. They also mention that a marketplace where proprietary data and modules can be sold could be a possible future for the industry.

And we couldn’t resist asking ChatGPT what it would take to clone ChatGPT. Here’s the answer ChatGPT gave:

The cost of cloning ChatGPT would depend on several factors, including the resources and infrastructure available to the organization or individual attempting to clone the model. Some of the main costs that would be involved in cloning ChatGPT would include:

1. Data: ChatGPT was trained on a large dataset of text, so acquiring or generating a similar dataset would likely be one of the biggest costs involved in cloning the model.
2. Computing power: Training a large language model like ChatGPT requires significant computing power, which would be another major cost.
3. Research and development: Developing a model that is similar to ChatGPT would require a significant investment in research and development to replicate the model architecture and fine-tune the model to perform well on specific tasks.
4. Human resources: The team that will be behind the development of the model would also be a cost, including data scientists, researchers, engineers, and more.

It’s hard to give an estimate for the cost of cloning ChatGPT as it depends on the size of the dataset and computational resources used, but it’s likely to be expensive and would require significant investment from a company or organization.

What are the odds ChatGPT will gain sustainable first-mover advantage?

Here’s how ChatGPT summarized the conversation:

In this conversation, the speakers are discussing the odds that OpenAI’s ChatGPT will gain sustainable competitive advantage by the end of the decade. They express varying levels of confidence, with one speaker giving odds of two to one, another giving odds of four to one, and a third giving odds of five to one to seven to one. They agree that OpenAI is currently the leader in the field and that the question is whether they can execute and maintain that lead. They also mention that competition, execution on the product side, and shifts in the developer landscape could impact the odds. They also mention that the company’s business model, which is going to pay back to the investors and give money back to the public, might have an impact on the odds.

Listen to John Furrier and Sarbjeet Johal debate the odds that OpenAI will sustain first mover advantage as we approach the end of the 2020s.

How did ChatGPT perform?

Thanks to ChatGPT, which provided a generally vanilla but well-written summary of the conversations about ChatGPT. Notably, the transcript used to feed ChatGPT was completely AI-driven. The results shown from ChatGPT are 100% verbatim. And despite the spelling errors and relatively subpar quality of the transcript, ChatGPT provided a nice summary of the conversation within each section.

However, while its summaries were crisp and easy to follow, the machine did not produce results that captured the nuance and more substantive points of the conversation. Please listen to the full video and you’ll see that the analysis provided by Furrier and Johal was much more analytical and specific.

Nonetheless, one can’t help but think back to the early days of the Netscape Navigator browser, something Furrier mentioned in our conversation. It was clunky, slow and quite ugly (below).

But it blew us all away and ushered in a durable paradigm shift that changed the world, in much the same way as iPhone.

Will conversational AI such as that demonstrated by ChatGPT have a similar effect on the world? Yes, we believe so. That natural language and conversational interfaces will completely change the way we live, work and play. Just as Amazon turned the data center into an API, this type of technology will turn technology and many tasks into a language command.

“Paint my house” is a possibility someday.

Will OpenAI be able to leverage its first-mover advantage to create a sustainable and dominant position? It’s anybody’s guess, but ours would be no… not likely. There are too many forces at play between the large internet giants, government oversight and massive investments that will plow into this domain.

The history of the technology industry suggests that a firm or group of companies will come from nowhere to disrupt and dominate. They will be rejected by dozens if not hundreds of VCs because they don’t match the safe pattern.

Perhaps you can be the one to crack the code. Either way, OpenAI has released the genie from the bottle and it’s not going back in.

Good luck navigating the future. As always we’ll be here reporting and analyzing it!

Keep in touch

Thanks to John Furrier and Sarbjeet Johal for the outstanding conversation. And special thanks to the Palo Alto studio team — Anderson, thanks for the cool backdrop. Noah, Chuck, Andrew, Cameron, thanks for letting us borrow your studio.

Many thanks to Alex Myerson and Ken Shifman on production, podcasts and media workflows for Breaking Analysis. Special thanks to Kristen Martin and Cheryl Knight, who help us keep our community informed and get the word out, and to Rob Hof, our editor in chief at SiliconANGLE.

Remember we publish each week on Wikibon and SiliconANGLE. These episodes are all available as podcasts wherever you listen.

Email david.vellante@siliconangle.com, DM @dvellante on Twitter and comment on our LinkedIn posts.

Also, check out this ETR Tutorial we created, which explains the spending methodology in more detail. Note: ETR is a separate company from Wikibon and SiliconANGLE. If you would like to cite or republish any of the company’s data, or inquire about its services, please contact ETR at legal@etr.ai.

Here’s the full video analysis:

All statements made regarding companies or securities are strictly beliefs, points of view and opinions held by SiliconANGLE Media, Enterprise Technology Research, other guests on theCUBE and guest writers. Such statements are not recommendations by these individuals to buy, sell or hold any security. The content presented does not constitute investment advice and should not be used as the basis for any investment decision. You and only you are responsible for your investment decisions.

Disclosure: Many of the companies cited in Breaking Analysis are sponsors of theCUBE and/or clients of Wikibon. None of these firms or other companies have any editorial control over or advanced viewing of what’s published in Breaking Analysis.

Image: Shafay/Adobe Stock