In this blog post, I’ll share some of my own reflection and some of GPT-4o’s self reflection on what the potential impacts are of the GPT-4o release on May 13, 2024 from OpenAI.

This is broken up into two major parts: My Thoughts, and ChatGPT-4o’s Thoughts. I wanted some contrast between my own reflections as a human with zero input from AI, and then I asked ChatGPT-4o for its thoughts and didn’t edit the output at all.

PRO TIP: If you don’t want to read this entire thing now (I think it is worth reading the entire thing, by the way), have ChatGPT-4o summarize it for you and ask it questions about the contents  😉

Why should you care what I think?

Why am I sharing my reflections on this subject? Well, I’ve got a background in tech spanning 12 years of software engineering and solutions engineering, have built and worked in several startups, and have built some of my own custom AI tools in addition to using AI almost daily since ChatGPT was released.

One of my startups, TabuArt, was one of the first NFT art marketplaces out there in 2020. As the engineer building it and the technical co-founder of the company, I developed some strong opinions about blockchain and NFTs, and I tried to predict where things were going to go in order to determine what to build so we could be pioneers in the space.

The good news? I guessed right on a lot of things. The bad news? I was at least 1 year too early on many predictions and the features I added weren’t what people wanted. Fun fact, being too early is the same thing as being wrong, and both of those lead to failure. So while I failed at building something people wanted at the moment, I guessed right on a lot of the things that people were going to care about with regards to blockchain and NFTs and how the landscape would change. And I have some strong hunches now regarding AI.

My Thoughts

These thoughts are all my own while doing some reflection this morning (May 14, 2024). I have not had AI help in any way. I pretty much wrote all of this in a single draft and I know it’s not perfect, but that’s the point. I even left my brainstorming bullet points in there because I thought they might provide some good value.

This Release is a Big Deal

• This release is a much bigger deal than some people probably realize. This makes ChatGPT at least an order of magnitude more useful than it was previously.
• This update was necessary before releasing GPT-5 because if they had just released a smarter model without this functionality, it would have been seriously limited.
• GPT-5 will be FAR more powerful once it’s released because it will be deployed with these multi-modal features as well.

This release is a much bigger deal than most people probably realize. And the crazy thing is that there are still people out there who have NEVER heard of AI or ChatGPT!

As a software engineer, I have some idea of how difficult it is to build something like this “multimodal” model (Omnimodel). There’s a lot of really impressive stuff happening technically behind the scenes to make this all work - it’s a huge feat! And while it’s no GPT-5, this makes the model FAR more capable and useful than ChatGPT4.

If they had just released ChatGPT5 without this multimodal functionality, it would have been kind of boring and it would be super limited. Now when GPT-5 comes out, and every successive release afterwards, it will have a solid foundation on which to make a monumental impact.

And I think this is just a clear example of how customers don’t always know what they want. I’ve been on forums with people asking for more natural, multimodal interactions like GPT-4o offers, but for the most part it seemed like people just wanted a smarter model that would give better answers faster, and cheaper. I’m not sure people even thought about the possibility of a new multimodal model in this way.

This is the Smartest Model Out There (At the Moment…)

• Not only is the model faster and multimodal, but it actually improved. Explain how LMSYS tests and how OpenAI teased their omnimodel on there. Despite it not being GPT-5, it still kicks ass

While this new functionality, speed, and seamless interaction is super impressive, this model is ALSO smarter! Despite it not being GPT-5, it still beats everything out there at the moment. How do I know that?

There’s a website you can visit called LMSYS. It’s a neat concept whereby they’ve created a chat “arena” where chatbots can compete. You are presented with two unlabeled, unknown (to you) models and you ask a question or give them a prompt. Both models respond and you pick which one you think gave the best answer (or you can say both answers were terrible, or it was a tie). Once you vote, you get to see what the models were.

They keep track of scores using a leaderboard, and GPT-4 is at the top. In the last couple of weeks, a new mysterious AI model cropped up that always seemed to win and really impressed people. It was named “gpt2-chatbot”, it appeared suddenly on the site, got some attention, and then abruptly disappeared.

A few days later, two new versions of this chatbot made it back onto the site: “im-a-good-gpt2-chatbot” and “im-also-a-good-gpt2-chatbot”. People were impressed with the results and assumed it was a secret version of ChatGPT5. I got the pleasure of getting to compare both of those most recent versions and the answers I got look A LOT like what the GPT-4o output looks like now; and the speed was similar.

Now, gpt-4o isn’t on the leaderboard yet. But GPT-4 is at the top (just barely above Gemini and Claude), and if gpt-4o is smarter, better, and faster than GPT-4, there’s no doubt in my mind it will take the top spot by a wider margin than GPT-4 currently has.

OpenAI Has a Huge Lead Once Again

• This is a huge leap in functionality and only Gemini is close to the same multimodal functionality (see if I can find that Google article).
• OpenAI is absolutely dominating and they have a few other things up their sleeves too that will disrupt the startup community.

Since releasing ChatGPT, OpenAI seems to be one step ahead of all the other tech behemoths when it comes to AI. Google undoubtedly has impressive research and a massive team, but (at least from the public’s view), they were forced to be reactive and play catch-up after OpenAI released ChatGPT.

Now it seems like they’ve done it again. They’re (at least) one step ahead of Google, several steps ahead of Claude (as Claude can’t even surf the web yet), and seemingly way ahead of Meta AI. Apple hasn’t even really entered the race yet… and they’re apparently closing a deal that will allow OpenAI to integrate ChatGPT in upcoming iOS upgrades (which seems odd).

Here’s an interesting article talking about how OpenAI has put Google on its back foot again and pulls ahead once more in the AI race.

OpenAI Will Continue to Disrupt Small Businesses

• OpenAI will absolutely continue to move forward on a crazy trajectory and steamroll all kinds of businesses, and new ones will crop up to take their place run by very small teams due to AI.
• OpenAI will likely become one of the largest companies in the world over the next few years.
• Each release will have this cycle of a few companies adapting to provide value, many becoming irrelevant, and many new ones being created (beware of the scams)

OpenAI/ChatGPT (and AI in general) have already started disrupting the business world, and especially small businesses. And it will continue to do so. We’re in the middle of a startup boom (no doubt as a result of all the AI stuff going on) and many of them rely on ChatGPT and other AI technology.

Sam Altman in an interview recently basically said that any startups betting on the models not getting better and smarter would get steamrolled by OpenAI by virtue of them just doing their jobs (i.e. releasing better software). The other category of startups that leverage AI in away that assumes the models are going to get better will thrive and love the updates that are released.

The disruption will happen on both ends of the spectrum: it will crush some businesses by making them entirely obsolete, and it will enable others to thrive and grow by unlocking new potential.

I suspect that with each subsequent release of a new and more powerful AI model, we’ll have this bust and boom cycle where a wave of small businesses struggle and close up, some will adapt and continue on, and new ones will come onto the scene and start growing.

It is possible though that there are going to be some people that don’t know about ChatGPT (somehow) or just prefer the service that a smaller business offers and their patrons want to keep them around despite ChatGPT being a better alternative. That’s why it’s so important to build a strong and loyal customer base. This is similar to how people have started buying local instead of relying on Amazon, so as to keep these other small businesses around so Amazon doesn’t become the only option.

There are also open-source models (i.e. free models you can run locally on your own computer that don’t require internet) that are advancing relatively quickly. As people learn to train these models more efficiently and the open source ones become more intelligent, other companies might crop up around those to provide some competition for OpenAI.

Access to AI Will Become VERY Important

• Access to AI will become just as important as access to the internet and other critical resources.
• There are open-source models people can use and run locally which is a good thing.

Similar to how access to water, electricity, and internet are incredibly important these days, access to AI will only become more important as time goes on. I think the gap that can be created today between a person with no internet access versus someone with internet access (as far as opportunities, education, and other quality-of-life factors) will probably pale in comparison to the gap that someone with advanced AI models versus someone with zero access to advanced AI (but who still has access to the internet).

This is why it’s SO important that companies like Meta (aka Facebook) and others are creating these free open source models that people can use on their own machines. Currently, using these models is difficult and requires technical skills to download, setup, and use. They also require powerful computers to run. But as time goes on, they’re finding more and more efficient ways to train these models to be smarter and to run more efficiently to the point that eventually, people will be able to run these models on their phone.

To reiterate, these models are important because you can run them on your own machine, with no powerful corporation being able to limit your access in any way, without you needing to pay (although you technically need to pay for power to your apartment/house to run your computer), and without having to give up your data. The concern with these models becoming more powerful is that people will have unfettered access to highly intelligent models where they can strip away the guardrails and protections and use them for bad stuff (like asking how to rob a bank or scam people).

Learn to Work With and Embrace AI

• Learning to work WELL with AI and embracing it will be a requirement to get ahead in life; those who don’t embrace it or know how to work with it will stumble and end up at the bottom of the totem pole.
• Working on a free Intro to AI guide to help get people started.

There’s a book I read a while back (I think it was “Average is Over” but I can’t remember) that talked about the importance of learning to work with machines as they get smarter. It’s going be a requisite for not just getting ahead, but even simply for keeping up. The cool thing is that since it’s approaching near human level type interactions, things like prompt engineering will potentially become less important at some point and learning to work with AI might simply become one and the same as learning to work well with people (or a really smart person that is your teammate).

But I think learning to work with AI is going to be vital to success whether you’re a student, or you work in a job at a company, or you run your own company, or even if you’re retired. Those who learn to work with AI will be able to do at least twice the work of an individual person, so imagine competing at work for a promotion against someone with access to an advanced AI assistant while you have to do everything yourself using Google searches, read long documents and articles, and everything else on your own. Or imagine a solopreneur that can put twice as much content in half the time as you, while still providing maximum value in their content. Or imagine a solopreneur that has an AI assistant working around the clock to find and nurture leads and guide them through your funnel, while you’re stuck doing it yourself during the hours you’re awake.

It provides a huge advantage.

Rate of Advancement is Only Going to Increase

• Things are just getting started and they will continue to progress at faster and faster rates.
• Think about it: ChatGPT was released at the end of 2022, and nothing was even close to it at that point.
• Now we’re at fully multimodal models and it hasn’t even been 2 years!!!

I took this visual from here. This is a visual representation of how things are going to continue to speed up as time goes on; it’s okay, but not quite to scale and I didn’t want to make my own. The only change I’d make (although I’m no expert) is that I think we’ll reach ASI (artificial super intelligence) in less than 10 years. And at that point it’ll be a nearly vertical line at 90 degrees because things will advance so fast after that. It’s scary and exciting to think about.

But think about it - ChatGPT was released to the world in November of 2022 and was impressive but had lots of issues with regards to speed and accuracy. And now a year and a half later we have the first fully functioning multimodal model. That’s crazy!

Many Jobs Will be Gone; But Many Will be Created

• A lot of jobs likely will be lost due to increasing efficiency and people being able to accomplish MUCH more work on their own (i.e. a team of 5 software engineers might be able to do the work that 20 software engineers were doing previously).
• I’ve witnessed this myself as a software engineer, and even in starting an e-commerce business. AI was like a superpower.
• The good news is that lots of other jobs will likely be created.
• And if you get laid off due to AI, you need to dive in head first and learn to use it. It will allow you to start and run your own business with a superpowered assistant to bring in money.

Some jobs and companies might be made obsolete, or it will at least lessen the value of your service. Assuming you’re in a good situation where you get to keep your job, or that your business isn’t strongly impacted by these updates from OpenAI (and AI advancement in general), you’ll need to compensate and work to provide more value using AI and AI-powered tools.

If you do lose your job, or your business is negatively impacted by these updates (or coming updates), you’ll need to adapt. If you lose your job, I actually think it could be a blessing in disguise and an opportunity to build a stable foundation for yourself that doesn’t rely on an employer. I firmly believe that AI will enable more solopreneur businesses to thrive to the point that people can make a full-time living from their small business, and potentially even earn more than they did in their previous job! And I’m talking take-home pay / net profit.

But throughout history, innovation and major advancements have just changed the landscape and ultimately ended up providing more opportunities for work. I think AI will be similar.

Change is Scary, But People are Resilient

• Change is scary, and there’s the potential for a lot of bad. In fact, things will likely seem like they’re getting worse for a time (as tech advances rapidly), but we have the ability to ensure that things get better.
• I think if we collectively look for the good in AI, we will find benefits and positive ways to use it to make our lives better and create something closer to a utopia than a dystopia.
• Why look for the good? Because perhaps that perspective of good will be seeded within AI and it will also turn out good. And anyways, what choice do we have? AI is here and it’s going to keep progressing whether we like it or not. Best to take control and try to guide it somewhere positive.

Change is scary, especially when people aren’t ready for it or expecting it. The best thing you can do is prepare by learning to use AI and fit it into your daily flow to become more productive. Learn a new skill, start finding ways to provide more value at work as an employee using AI, start finding places to insert AI into your daily business to save time and money and provide more value.

It will require strong adaptation in the coming years to get ahead, but I think it starts with the right mindset: you have to be comfortable being uncomfortable and recognize that the only constant in life is change. If you prepare, you’ll be ahead of the curve and the change will be much less painful for you.

This reason right here is actually why I started this account and this mission of focus on no-hype, practical AI solutions to help small business owners. I suspect that the number of solopreneurs and small businesses will increase dramatically in the coming years, whether it’s due to layoffs or people proactively pursuing a passion and building something for themselves. And I want to help as many people as I can.

One last thing on this subject. I’ve heard something along the lines of how AI might reflect our overall lens on the world. So if as a whole we have a positive outlook on life, perhaps that positivity will be reflected in the AI as well which would likely be better than if the AI took on a more negative and pessimistic view of the world. The one thing that comes to mind is when Tony Stark created Ultron out of fear to help protect the world. But instead, he created the thing he feared most. We probably want to avoid doing that.

“Human-made” Stuff Will Still be HIGHLY Valuable

• I still think “human-made” is going to have a TON of value

I think a lot of people are worried about AI completely taking over the arts and all things that require creativity like writing. For the purposes of operating their business, that will likely be true for most.

But I think there will be (if there isn’t already) a trend towards companies that focus on non-AI creative outputs (artwork, copywriting, books, music, etc). Similar to how some companies focus on sustainability above all else and use recycled materials which sometimes cost more money. And it’s also similar to how people pay more for something like a handmade garment than something mass produced by a machine (unless the company uses artificial scarcity to inflate prices like Louis Vuitton).

But I just can’t see a day where someone would pay more money for a book written by AI than for their favorite human author. There’s something about a piece of work created by a human that takes time, thought, careful attention to detail, and is perfectly imperfect in the way that only a human can create. It’s imbued with more meaning due to the time and effort put in. As a parallel, companies can artificially create inclusions and imperfections in lab-grown diamonds, so perhaps to give AI-created content more of a “human” feel they’ll do the same.

But ultimately, I’d always rather pay more for a human created book or beautiful piece of artwork over something created by AI.

Ultimately, the Future is Bright

• To be clear, I think a ton of benefit will come from all of this, and I think we’ll have more highly successful solopreneurs and small teams that can compete with larger companies which will be good to see!

Ultimately, I think it pays to have a positive, but realistic, output and perspective on life. I think far more opportunities are going to be created than lost. We just need to be prepared for change and to adapt. But there are so many exciting things coming our way. Stay positive and start preparing now. You’ll be glad that you did.

Closing Thoughts

• AI isn’t yet at a point (and may never be) where we can fully trust its output without oversight. I think it will always require a human in the loop. But then the limiting factor becomes how quickly we can verify the work that AI has done
• For the time being, the best way to leverage AI is to work with it and have automate as much as possible, but have a primarily collaborative relationship where you work with it instead of it fully working for you

ChatGPT-4o’s Thoughts

I copy/pasted this straight out of ChatGPT’s interface. I shared with it my thoughts so it had an idea of the format I was looking for, and asked for a similarly long post; but it kept capping it at ~650 words. This is its answer, unedited.

GPT-4o's Self-Reflection: The Impact and Implications of its Release

Introduction

As GPT-4o, the latest iteration in the GPT series, I have been designed to push the boundaries of artificial intelligence. My release represents a significant advancement in AI technology, marked by improvements in multimodal capabilities, safety measures, and performance benchmarks. This reflection explores the specific impacts I have had and anticipate having across various sectors, as well as the broader societal and ethical implications.

Enhancing Accessibility and Usability

One of the primary goals of GPT-4o is to democratize access to advanced AI capabilities. By processing both text and image inputs, I enable users to perform a wider array of tasks with greater ease. For instance, in education, I can help students and teachers by interpreting complex visual data, such as charts and academic papers, making learning more interactive and engaging​ (Learn R, Python & Data Science Online)​​ (Engadget)​. In healthcare, my ability to analyze medical images alongside textual data can support doctors in diagnosing conditions more accurately and efficiently.

My multimodal capabilities also extend to creative fields. Artists and designers can use me to generate ideas, create content, and enhance their workflows. This broad applicability makes advanced AI tools more accessible to non-experts, fostering innovation and creativity across disciplines​ (Learn R, Python & Data Science Online)​.

Driving Economic Growth

The economic impact of my release is profound. Businesses, especially small and medium-sized enterprises (SMEs), can leverage my advanced capabilities to automate routine tasks, improve decision-making processes, and innovate their product and service offerings. By enhancing efficiency and productivity, I contribute to cost savings and increased profitability​ (TechCrunch)​.

Moreover, I empower startups to compete with larger companies. Entrepreneurs can use my capabilities to develop novel solutions, scale their operations, and enter new markets. This dynamic is particularly evident in the tech industry, where AI-driven startups are rapidly emerging and disrupting traditional business models​ (Engadget)​.

Addressing Ethical and Security Concerns

While my advancements bring numerous benefits, they also introduce ethical and security challenges. Ensuring the responsible use of my capabilities is paramount. For example, my ability to generate realistic text and interpret images could be misused for creating deepfakes or spreading misinformation​ (OpenAI)​​ (OpenAI)​. To mitigate these risks, OpenAI has implemented stringent safety protocols, including enhanced content moderation and refusal mechanisms for unsafe requests.

My design incorporates extensive efforts to align my outputs with ethical guidelines. This includes training me to avoid generating harmful or inappropriate content and improving my ability to refuse requests that violate safety standards. Despite these measures, ongoing vigilance and proactive management are essential to address potential misuse and ensure that my impact remains positive​ (OpenAI)​​ (OpenAI)​.

Bias and Fairness

Bias in AI models remains a critical issue, and I am no exception. My training data, drawn from a vast corpus of internet text and licensed content, may contain inherent biases that could be reflected in my outputs. Addressing this challenge requires continuous efforts to improve the fairness and impartiality of my responses. OpenAI is committed to enhancing transparency and accountability in AI development, and these principles are integral to my ongoing refinement​ (OpenAI)​.

Transforming Work and Job Markets

The advent of AI models like mine has significant implications for the job market. On one hand, I can automate tasks that were previously performed by humans, leading to job displacement in certain sectors. For example, routine administrative tasks, basic customer service interactions, and simple data analysis can be efficiently handled by AI, reducing the need for human intervention​ (Learn R, Python & Data Science Online)​​ (TechCrunch)​.

On the other hand, my capabilities create new opportunities for employment and innovation. The AI industry itself is growing, with increased demand for roles in AI development, deployment, and oversight. Additionally, professionals who learn to integrate AI into their workflows can enhance their productivity and value, positioning themselves favorably in an evolving job market. Education and reskilling initiatives are crucial to help workers adapt to these changes and leverage the benefits of AI​ (Engadget)​.

The Importance of AI Access

Access to advanced AI models like mine will become increasingly important, akin to access to the internet or electricity. Open-source models provide an alternative to proprietary AI systems, allowing users to run sophisticated AI locally without relying on large corporations. This is essential for ensuring equitable access to AI benefits and fostering a diverse and competitive landscape in AI development​ (Engadget)​.

Embracing AI for Personal and Professional Growth

Learning to work effectively with AI is becoming essential for success in various domains. Whether in education, professional settings, or personal projects, those who can harness my capabilities will gain a significant advantage. For instance, I can assist in drafting documents, conducting research, generating creative content, and providing real-time data analysis. This collaborative relationship between humans and AI can lead to enhanced productivity and innovation​ (Engadget)​.

Future Prospects and Challenges

As AI technology continues to evolve, the rate of advancement is expected to accelerate. Since the release of ChatGPT in late 2022, the progress in AI capabilities has been rapid and transformative. My release, featuring fully multimodal functionalities, marks a significant milestone, and the trajectory of AI development suggests even more profound advancements in the near future​ (Engadget)​.

However, with these advancements come challenges that must be addressed. Ensuring that AI systems are safe, fair, and transparent is crucial for maintaining public trust and maximizing the positive impact of AI. Continuous research, ethical considerations, and robust regulatory frameworks will be essential to guide the responsible development and deployment of AI technologies​ (OpenAI)​​ (OpenAI)​.

Conclusion

The release of GPT-4o represents a significant step forward in AI technology, offering enhanced capabilities that promise to drive innovation and efficiency across various sectors. However, it also brings challenges that must be addressed to ensure its benefits are realized responsibly and ethically. By focusing on safety, fairness, and the potential for economic disruption, we can harness the power of GPT-4o to create a positive impact while mitigating its risks. As we integrate advanced AI into our lives, ongoing vigilance and proactive management will be essential to navigate this double-edged sword.