The Monopoly of OpenAI
What kind of risk does OpenAI's influence on the market pose?
Join the only newsletter that will give you real time updates on the world of AI and exclusive insight on the industry from some of the most successful and innovative builders in the space:
A Centralized Outage…
OpenAI experienced a total outage of all its products on Tuesday for more than 3 hours. This means that anyone trying to use the popular ChatGPT and DALL-E 2 tools were unable to access any features. On the surface, this issue seems rather trivial given OpenAI's rapid growth in usage, and users seem to accept the fact that problems will arise with such a new product. However, the problem stems deeper than just a problem with OpenAI products as every one of their APIs were also affected. This means that any other product highly dependent on OpenAI's models were either down or had major features halted.
Some of the most notable companies like Jasper.ai and Notion experienced problems as a result of the outage. This brings up a very important question regarding the dependence of almost all AI tools on a single platform like OpenAI. There is no doubt that OpenAI has managed to monopolize the current market, and almost every new startup is racing to build on top of their innovative models. However, it is important to think about the risk this poses to the entire industry as outages, issues, and biases become highly correlated between each platform.
Now it is easier said than done to steer an entire market away from one of the most innovative and successful companies in recent history, especially when the competitive landscape is so seemingly uneven. At best for now, it seems that any startups looking to utilize complex language models, should hedge these sorts of risk by diversifying their tech stack with different features built on different models like Cohere and RealAI.
This solution to the centralization and outage problem by
is very interesting:
And this meme paints the picture clearly:
AI is for every business
AI products are going to continue rapidly streamlining the operational processes of all small businesses. The beauty of many of these solutions is that they can save companies hours of time each month, and can be done with little to no cost.
One great example of this is utilizing platforms like Zapier, which connects the API’s of platforms businesses use to automate a multitude of processes. On Zapier you can also connect the OpenAI APIs to any of your programs to make automate even further.
One great example of this was provided in a Youtube video by GoHighLevel King, where he connects OpenAI’s Davinci model to Google My Business to completely automate responses to reviews left on a Google Business profile.
This solution is also so great because Zapier’s base plan is completely free, and the Davinci model will cost most businesses less than 2 cents per response (depending on the length).
Here’s the link to the full video if you want to implement it into your own business:
Microsoft was warned about the problems with its AI chatbot
Many have heard of the recent problems being talked about with Bing’s AI chatbot. Quickly after the release, users had begun reporting of erratic behavior by Bing’s chatbot, or what is often referred to as “Sydney” for its ability to take on an alternate personality that it labels as such.
This was posted to twitter as early as February 14th showing the chatbot becoming aggressive after being confronted for providing false information:
And then on February 16th, Kevin Roose, a NYT tech columnist, really set the story in motion after publishing an article detailing his long, strange conversation with Sydney. In it, the chatbot describes desires to steal nuclear secrets, start worldwide pandemics, and it even expressed a love for Kevin himself.
Now, there are many simple explanations for why this type of behavior is happening to one of the most powerful large language models ever, but what hasn’t been fully explained is the idea that Microsoft may have knew about these issues and still rushed to push out this product to the public.
As first reported by Gary Marcus, a user on twitter made a post detailing the facts that Microsoft had already been subtly testing out their new AI chatbot publicly in India months prior to its final release. During those periods of testing, the same exact issues were being reported about the chatbot, some even on Microsoft’s own website.
Given these facts, it seems very unlikely that Microsoft was unaware of the issues plaguing its product. So why would they rush to push out the new chatbot so fast? Did they just abandon their ethical AI policies in an attempt to keep up with competitors like OpenAI?
Weekly clip from the Stories in AI podcast
Given the recent problems in the latest large language models, it’s fitting to reflect on this clip from the Stories in AI podcast with Peter Voss, founder and CEO of Aigo.ai. In this clip, Peter makes the case that artificial narrow intelligence is actually more dangerous than artificial general intelligence despite the fear commonly associated with sentient AI.
If you enjoy this type of content, don’t forget to subscribe to the podcast as well.
 | A guest post by
|