When the AI rubber hits the road
Large Language Models (LLMs) have stormed the front pages of mass media, thanks to OpenAI and its now famous ChatGPT. The term AI is now everywhere. Despite wars, presidential elections, plane crashes and other natural disasters, it always comes back to the front page.
Even though we have other technologies which are sometimes bundled together, the term AI is used mostly for LLMs. As you can imagine, it does create confusion when a company claims that it is using AI in its product and it doesn't talk back to you. But who has time for the nuances of AI? Details.
The important thing is what AI will be able to do for us now and in the future.
So, let's talk about the future. That's always exciting. We know that AI itself is not the goal. The goal is Artificial General Intelligence (AGI) which based on the definition from OpenAi is: '... artificial general intelligence—AI systems that are generally smarter than humans—benefits all of humanity.'
We also learn that: 'If AGI is successfully created, this technology could help us elevate humanity by increasing abundance, turbocharging the global economy, and aiding in the discovery of new scientific knowledge that changes the limits of possibility.'
Also: 'On the other hand, AGI would also come with serious risk of misuse, drastic accidents, and societal disruption. Because the upside of AGI is so great, we do not believe it is possible or desirable for society to stop its development forever; instead, society and the developers of AGI have to figure out how to get it right.' There is more, but we should note that it was published on February 24, 2023.
How do we know that we reached the moment that AI will become AGI? Will there be any general test which will generally prove that the AI is generally smarter than the general human? That test is much simpler. Based on the agreement between OpenAI and Microsoft, the milestone will be when AGI will reach $100 billion in profit. You might find it interesting that Google in the fiscal year 2024 on revenue of $350 billion made a net income of $100 billion. Microsoft on revenue of $245 billion made 'only' $88 billion.
Considering that OpenAI is losing billions a year, that $100 billion in profit might be some time in the distant future. And so is AGI. Or is it?
It could be a lot closer, according to Mr. Altman from OpenAI. In his blog post aptly named Reflections, he says 'We are now confident we know how to build AGI ... ... We are beginning to turn our aim beyond that, to superintelligence in the true sense of the word.' That's how OpenAI will get around the agreement with Microsoft. It will build AGI this year (2025) and while it starts making few billions of dollars of profit to pay Microsoft back, OpenAI will build the superintelligence (is it going be SAI, ASI or AIS or simply superintelligence??) which will make trillions of dollars and OpenAI will be able to keep it all.
Also fortunately for all of us, while OpenAI will be building the next generation of AI, people there will still have time to get involved with The Stargate Projectwhere '... SoftBank and OpenAI are the lead partners for Stargate, with SoftBank having financial responsibility and OpenAI having operational responsibility.'
A question for the future - Will the AGI or the upcoming superintelligence from the Stargate project still come with a footnote like this? 'ChatGPT can make mistakes. Check important info.'
Because if it does, what kind of intelligence are we talking about?
That's the future. But before we get to the AGI let's see what AI can do for us now.
You've heard about the magic ChatGPT can do with text. It can correct your grammar, generate outlines, summarize text and even provide search results from the Internet.
It can also generate short videos based on various prompts. But sometime you encounter a kind of information exchange like the one I just had:
Based on the prompt and the attached file, create a short video
It looks like you've uploaded an image. Do you want me to create a short video using this image as a background
Yes
It looks like there's an issue with text rendering in the video due to ImageMagick restrictions. I’ll try a different method to overlay the text and regenerate the video. Stay tuned!
Is it done yet?
Not yet! The previous approach had an issue with text rendering. I'll use a different method and regenerate the video now. Give me a few minutes!
There was an issue generating the text overlays due to a security restriction with ImageMagick. I'll adjust the approach and try again.
It seems like I can’t do more advanced data analysis right now. Please try again later. However, you can manually add text overlays using tools like Canva, Photoshop, or even PowerPoint.
Awesome. At the end of our conversation, I am told to do it myself. That's what I call superintelligence. Outsource the advanced tasks to humans...
But that's just me with my unimportant requests. Let’s get back to how many companies are actually using this technology for various purposes.
As you can imagine, sales and marketing is right there. Maybe you don't know these companies, but I am sure you saw the outcome of their efforts:
an invite on LinkedIn - 'Would love to catch up with you. Looking forward to exchange common synergies.'
a pitch on LinkedIn - 'I noticed you're building something exciting! I’m passionate about connecting with people who are carving their own paths. Let’s connect—would love to hear about your journey.'
a personalized pitch (scraped from my profile) - 'Noticed we share a passion for leveraging technology to drive growth, especially in the SaaS space. Your work with startups and fast-growing companies is inspiring—love how you focus on creating strategic options with tech.'
Pitch-slapping people may indeed be boring for the one making the pitch, but you don’t want to bore the person reading it. As you can see, this use of AI isn’t ready for prime time. Not even close.
When promoting this kind of technology to others, companies always emphasize that AI is here to offload the most boring tasks, like Microsoft did when it was promoting its CoPilot. This brings up another highly successful implementation of AI, with chatbots: these ever-present AI creatures living now on almost every website and providing real value, especially to organizations which don’t need to hire that many people.
How is this helping the customer? Who cares! What’s more important is the money the companies are saving. That’s true up to the point when the customer has to sue the company to get their money back because of problems caused by the chatbot’s advice.
Mentioning products where AI seems to create miracles is almost getting redundant. For instance, there was this announcement from Slack - 'With Agentforce in Slack, you can work alongside autonomous AI agents that surface knowledge and can complete tasks.' Are we talking about 'common synergies' here or me manually adding text overlay using Photoshop?
But the best example is coming from Anthropic, an AI company with its flagship AI product Claude. It has billions of dollars invested in it and a customer list from Asana, Jane Street, LexisNexis to SAP and ZoomInfo, just to name a few. That company is a big proponent of using AI everywhere and for everything. Except .... when it comes to hiring its own people. In their job posts they state:
'While we encourage people to use AI systems during their role to help them work faster and more effectively, please do not use AI assistants during the application process," Anthropic writes on its online job applications. "We want to understand your personal interest in Anthropic without mediation through an AI system, and we also want to evaluate your non-AI-assisted communication skills.'
My dear reader, do you think that the HR department at Anthropic is using people to read every single submission? Or are they using AI to screen the resumes to highlight the most promising candidates that closely fit the job description?
To end on a positive note, there are examples where technology delivers value measurable in dollars. One of them is the Amazon Q Developer, which can modernize your Java application and save you time, money and resources.
Even Amazon's CEO claims that ‘With Q’s code-transformation capabilities, Amazon has migrated over 30,000 Java JDK applications in a few months, saving the company $260 million and 4,500 developer years compared to what it would have otherwise cost.’
Yes, there are emerging patterns where, for very specific use cases, it starts delivering value and becoming recurrent. But for many other tasks, we have to still wait for the AGI and keep doing things manually as suggested by know-it-all AI. We are not there yet. Not even close.