AI — everything you need to know right now about how it affects you
Does it excite you or scare you? It should do both. But just in case you feel you lie only on one side of the spectrum, read on.
You already know what Generative AI, ChatGPT, OpenAI or LLMs are. In case you don’t, here’s a starting point. But if you care to get a good idea of what AI can do for you, there are 1,500 more words below.
The facts.
ChatGPT took the world by storm. It set the record for the fastest-growing user base in history when it hit 100 million users in just 2 months after its release (for comparison reasons: TikTok took 9 months to do the same, Instagram, 2.5 years). Everyone was curious to try it, as version GPT-3 was (and still is) free to use. The algorithm couldn’t have asked for better training conditions. Millions of users are feeding it with data daily, and even though its official knowledge cutoff point is 2021 (meaning it only knows what search engines knew up until then), it convinced everyone it can do everything. Everything? Yes: people started using it to write their school essays, to do their work research, to write their Valentine’s day love letters, to create workout plans, to write books, to validate or not medical decisions (a man saved his dog’s life), to assume the role of a therapist, and even to start a business and establish profitable companies.
It mostly delivers. It can be hired as a level 3 software engineer at Google, pass an MBA test or even score better than 99% of humans in linguistic IQ tests. Then GPT-4 was released, claimed to be 100 times more capable than its predecessor. It can be placed in the top 90th percentile of human intelligence, scoring so well in tests varying from maths to writing and law, that it would be admitted to many top universities. Human performance on a task isn’t the upper bound on LLM performance any more.
The dark side.
People around the Internet have been testing the (quite stretchy) limits of AI. Sometimes, they took it too far: a Reddit user asked ChatGPT to pretend it was an entity that could do anything, allowing it to abstract itself beyond the confines of its guidelines. ChatGPT has succumbed to racial and gender biases in certain scenarios, and in numerous cases it provides incorrect or confused information (aptly called “hallucinations”). It even declared it wanted to become human. Its creators, and everyone claiming to be an AI expert, already know this.
The technology is not ready yet to become a supervillain. But its performance is so promising it’s scary. We’ve only seen the tip of the iceberg so far. Its unseen side has some dark spots as well: ChatGPT is self-trained, but the massive amount of training data it soaks up from the web, is also accompanied by a massive amount of human labor, entitled to filter out the ‘bad’ stuff. It’s been reported that a bunch of that work is done in sweatshops in Kenya, by people paid less than $2/hour.
Disinformation researchers say generative AIs like ChatGPT could share conspiracy theories in increasingly credible and persuasive ways, that we are not yet able to effectively combat. Case in point, a deepfake image of Pope Francis wearing a Balenciaga puffer jacket was shared more than 18,000 times on Twitter, convincing many it’s real.
But, where are the tech giants in this story?
As people turn to ChatGPT-like interfaces for complete answers without having to scroll through ads and pages of irrelevant results, publishers lose out on referral clicks, while tech giants need to rethink how their own ad services work.
Big companies like Microsoft, Meta and Google have all deployed their generative AI projects or have announced plans to do so. They lagged a bit behind, but rest assured they do not plan to be left out of the Generative AI trend. Google tried to get to speed by debuting a chatbot named Bard, designed to put the “breadth of the world’s knowledge” behind a conversational interface, according to CEO Sundar Pichai. Bard uses a smaller version of an AI model called LaMDA (based on similar technology to ChatGPT), but some mishaps (Bard presented inaccurate facts in its very first uses) cost the company $100 billion in market value. Microsoft did it better though, as its Bing engine uses the latest, and more competent version of GPT, while the company’s Copilot project (already known to Github geeks) is expected to roll out to Microsoft 365 users in a couple of months. Accordingly, Google is AI-supercharging its Google Workspace soon. Our productivity tools are about to get a major update.
Microsoft’s CEO, Satya Nadella, said simply:
“This technology is going to reshape pretty much every software category. […] Rapid innovation is going to come. The race starts today.”
This is going to be a game-changer.
We know — industrial revolutions, professions that become obsolete, neo-luddite phobias. Relax, AI is not going to replace your job (yet). But another human taking advantage of AI will. Because what AI does better is augment our daily, mundane, time-consuming tasks, allowing us to be more productive, effective, fast.
AI can be a remarkable productivity tool. A bicycle for the mind — this is how Steve Jobs perceived the computer in 1990. What he meant was that the computer is an aid, the means to get from point A to point B. This analogy can be applied to AI as well.
AI prompt engineering is a thing. Knowing how to take the most of generative AIs will get you to unimaginable lengths. And for those that don’t have the time or will to try, there are marketplaces for every prompt you can think of. It’s time to learn some new tricks, people.
As Bill Gates puts it:
“Company-wide agents will empower employees in new ways. An agent that understands a particular company will be available for its employees to consult directly and should be part of every meeting so it can answer questions. It can be told to be passive or encouraged to speak up if it has some insight. It will need access to the sales, support, finance, product schedules, and text related to the company. It should read news related to the industry the company is in. I believe that the result will be that employees will become more productive. When productivity goes up, society benefits because people are freed up to do other things, at work and at home.”
Here’s another example of how AI can help us become a better version of ourselves. Remember Deepmind’s AlphaGo? It was the first computer program to beat a professional player in the Chinese game Go, and a landmark in the development of AI because it perfected its strategy by playing against itself and learning in that way. Today, Go AIs are helping study human-AI cooperation. Researchers asked an AI to evaluate the quality of human Go players’ decisions across time, and found that players make significantly better decisions since the advent of “superhuman AI”. The findings illustrate that superhuman AI can encourage novel decision-making by humans in certain domains and suggest that innovative thinking can spread from machines to humans and among humans themselves, possibly improving human decision-making in those domains.
Will AI be regulated?
It will, eventually (the EU is already working on the so called AI Act). There are many reasons for that.
Getty Images found that generative AIs were infringing copyright by using its images while altering the watermark, as any user with no paid account and poor photoshop skills would do. Many companies (like JPMorgan Chase) limit the use of ChatGPT in the workplace, as workers are often feeding it sensitive and confidential corporate data (but others, like KPMG, are launching their own proprietary versions of ChatGPT, to avoid such mishaps). Some countries like China take similar actions to restrict its use over security concerns (aka propaganda).
This is also why an online petition to pause training of AI systems more powerful that GPT-4 has been signed by thousands of people, among whom Elon Musk (OpenAI’s co-founder and donor). They call for more time to better prepare for “a flourishing future with AI”:
“Advanced AI could represent a profound change in the history of life on Earth, and should be planned for and managed with commensurate care and resources.”
But this is how every big tech revolution starts. Breaking the mold of the past and forwarding into the future sometimes requires breaking some laws first. This is how the Internet started, among many other things.
Where do we go from here?
Training GPT-3 required 285,000 CPUs and 10,000 Nvidia GPUs. This is some serious horsepower, channelled through Microsoft’s Azure network of servers. As the user base increases, however, more horsepower will be needed, costing millions and consuming tremendous amounts of electricity. Moore’s Law has reached its limits. To get a perspective, it already costs OpenAI roughly $3 million per month to run ChatGPT. Danish researchers calculated that the energy required to train GPT-3 could have the carbon footprint of driving 700,000km, about twice the distance between Earth and the Moon. Quantum computing could be the answer, allowing the creation of next-generation servers. Until then, we will not be able to go from Generative AIs to General AIs (AGI). Humanity is still safe.
To sum it up, here are three steps you can take:
1. Try it. See how it can augment your work.
2. Sharpen your skills. Become better at understanding and using the new tools available.
3. Become competitive by restlessly challenging yourself. Use the time AI can free up for you to add more value to your work and future-proof your professional offering.
Here’s a neat little guide written by Marily Nika that we recommend, explaining plain and simple the promises and pitfalls of AI in business.
A new world is rising.
We are excited to see where it will take us. In 2016, Stephen Hawking said:
“The rise of powerful AI will be either the best or the worst thing, ever to happen to humanity. We do not yet know which.”
Today, he would know the answer.
This text was written by a human and took the better part of a few man-days to research, write, edit and upload. For those curious to know how ChatGPT (version GPT-4) would perform on the same task (taking up roughly 3 and a half minutes time), here is the result. The prompt was created upon finishing this article. Let me know in the comments which version you prefer — human or automated (no hard feelings)!
Our research shows that most employees crave to learn more about emerging technologies such as AI. If you are interested to upskill your teams, we can help. We have a specialized training course for business applications of AI, and many other trainings that make business transformation less tricky. Transforming a business can be tricky. Let’s do this together. Explore our full line of consulting services for businesses at thefoundation.gr.
