Very recently OpenAI introduced their text to video generator, SORA. It was released without a lot of fanfare but timed to compete with Google’s announcement for its Gemini 1.5 large language model tool which rivals ChatGPT. It has enhanced efficiency and scale over the previous 1.0 release. SORA won the news cycle, however. The videos it generates are extremely high quality and realistic in terms of visual content. They don’t yet incorporate sound or music and the videos are limited to one minute. What is most impressive is that it demonstrates the ability to learn without direct assistance from scientists. It creates virtual realities with an understanding of the physics of our reality that it was not trained on. This knowledge is self-learned. This is what has people who follow the subject matter of Artificial General Intelligence (AGI) so impressed. AGI is the theory that AI can learn to solve any number of hypothetical tasks using generalized human cognitive abilities. It was believed that we were several years away from this accomplishment. It is possible that this is the reason for Sam Altman, OpenAI’s CEO, being fired from OpenAI—albeit for a day? It was reported that the board did not feel he was being transparent about the progress and development of AGI at the company.
Estimates were that we were seven to eight years away from such an achievement. Estimations of that timeline have now shrunk. We are most likely even further along the AGI path now than these incremental releases suggest once you factor in the time required for research, training, and development. The growth is proving to be exponential. OpenAI has other projects in the works which by the time they are released to the public they are already old, like light from a dying star reaching the Earth light years later.
OpenAI has stated that they are going to incrementally roll out ChatGPT-5 because they believe people will be overwhelmed by its intelligence. The current ChatGPT 3.5 has a verbal IQ score of 155, Einstein’s IQ was 165. ChatGPT-5 is sure to score higher than ChatGPT 3.5. The incremental roll outs are akin to boiling the proverbial frog in a pot. The frog gets used to the heat and by the time it realizes the pot is boiling it is too late to do anything about it. OpenAI recognizes that most of society is not ready for its tools, but it is in a race to beat all competitors so they must keep their edge and continue development.
Social Tensions
This year there has been a spate of driverless cars attacked and set on fire in San Francisco as people express their concerns about pedestrian safety and loss of employment due to AI. The introduction of AI is stoking social tensions already and we have not reached AGI yet. Even the developers of this new technology realize that the public is not “ready” and are releasing their tools incrementally and monitoring the response of the people before rolling out new tech hiding behind the curtain. Unfortunately, with the speed at which technology is being developed and the slow pace of legislation, it looks like the people and AI are in for a bumpy ride.
Disrupting Employment
Many companies are either replacing their work force with AI technology or studying how to do it. UPS recently announced a layoff of 12,000 of their 85,000 management employees. They are leveraging AI technology to replace the laid off workers to make the company more efficient and profitable. Some news organizations have spun the story suggesting that the recently negotiated contract with the drivers is responsible for the layoff numbers. I suspect that there is little truth in that. The managers are being laid off, not the blue-collar driving and warehouse union workers. They dodged the bullet this time around.
The producer, director, actor, Tyler Perry, has halted a plan that was four years in development to expand his current studios in GA at the cost of $800 million after seeing what can be accomplished with SORA. He notes that the tool will put a lot of film industry people out of work: lighting, grippers, writers, actors, location scouts, etc. Disney has also established a task force to study how their company can cut production costs. This could potentially put many animators, special effects artists, video crews, writers, and other creatives out of work.
Environmental Impact
There isn’t much discussion about the impact that large scale AI projects have on the environment. According to an article written by Kate Crawford in the publication, Nature, training large language models like ChatGPT 3 can consume over 400 watts of power while operating. Typically, you need to consume a similar amount of power for cooling and power management as well. Overall, this can lead to up to 10 gigawatt-hour (GWh) power consumption to train a single large language model. This is on average roughly equivalent to the yearly electricity consumption of over 1,000 U.S. households. This massive amount of electricity is not always generated by renewables. She notes that “It’s estimated that a search driven by generative AI uses four to five times the energy of a conventional web search. Within years, large AI systems are likely to need as much energy as entire nations.” U.S. legislators have introduced a bill directed towards this issue, The Artificial Intelligence Environmental Impacts Act, but it is not going to be adequate in its current form to address future issues.
Advanced Chip Production
Sam Altman is looking for investors to the tune of 7 trillion dollars to scale up AI chip production. He wants OpenAI to create their own chips so that production can keep up with demand and the machines can be even more powerful and fast in computational supremacy. Seven trillion is more than the amount of money it takes to fund the U.S. Government. He is looking to Saudi Arabia to become major investors. Such foreign investments may not necessarily be aligned with the goals of the recent Chips Act which is meant to give the U.S. an edge in domestic chip development. With so much money coming from foreign investors it could call into question issues of ownership and influence. Of course, it would depend on the number of shares the investor holds and how the deal is structured. This is not going to happen tomorrow, but it is being given very serious consideration.
I Smell Smoke
We are still barely able to smell the smoke and the house is on fire. The disruption of AI will be different from any other change to industry we have experienced because it will impact many industries at once or at least in a very rapid domino effect. It won't be limited to a single sector of the economy like say, just car manufacturers and the ancillary businesses related to them. We are talking about major changes to film, music, design, health care, legal, accounting, manufacturing, retail, call centers, and IT. Millions of people whom these businesses currently employ will be out of work at roughly the same time. They can’t just go and find another job when AI has replaced so many workers. We can’t treat the development of AI the way we would treat a company that manufactures a fancy new widget. The product has global influence and the potential to be incredibly disruptive.
Pausing Advanced AI Development
In an earlier piece I wrote about the discussion to pause advanced AI development. That does not seem to be happening and many in the field believe it is not even possible now. AI development has become a kind of arms race. Companies must establish their dominance. They don’t dare slow down for fear of losing relevance. These tools should benefit humanity and not just the CEOs, shareholders and employees of the companies that develop them. A handful of companies can decide who will be allowed to benefit from the technology developed, picking winners and losers. The public must weigh in on this very critical development. There is too much at stake and the public will be asked to bear the brunt of any failures or mistakes.
Sam Altman, Tim Rice, Bill Gates, Sundar Pichai, Mark Zuckerberg, Jeff Bezos, etc. are allowed to decide what is best for the world that we all live in. These men may be smart and have a talent for making money, but wisdom and intelligence are not synonymous and a person that claims to be a leader should have both. They are human beings like the rest of us. No better than anyone else due to their talents. Society may covet their gifts, but that does not make them superior human beings. Our technology must come from a place of compassion not hubris or it will not serve us well. It may in fact destroy us.
The Balancing Act
Our technology may be 21st century but our social and economic systems are based in the 20th century, and the two won’t play well together. We still cling to outdated ideas of productivity, labor, social welfare, and finance. It is obvious that AI is already disrupting these structures. It can’t support our old model of trading your labor to a company in exchange for monetary compensation allowing you to exchange that for goods and services. How do you keep a consumer economy going without enough paid workers who will spend the money that they make? The wealth gap will widen even more. We have been trained to define our worth through the work we perform. Without a way to fill this gap people may become very confused and depressed, leading to self-medication with drugs, exasperating a problem we already have.
Education, Immigration
AI may benefit those that embrace it, but how many people will that be? Industries like entertainment that employs a little over 4.4 million people as of 2022. Their head count can be reduced to a hand full of people writing prompts for the AI. There will still be a shrinking pool of jobs that will dry up eventually and won’t be coming back. What will the newly unemployed do for money when job opportunities are being shut down across the board? Creatives, white-collar and educated workers previously protected by their status and positions will be destabilized. We will have to reconsider the value and definition of education. Reeducation without some kind of assistance is going to be expensive and may not even be an option for some people. What would be the benefit of studying for a profession that AI may be doing in the very near future? ChatGPT has already passed the BAR and Medical Licensing exams.
Adoption of AI may curtail immigration to this country. There won’t be jobs available to attract people seeking a better life in this country. People that decide to come the country anyway to compete for works may really tick up the social tensions around AI. If immigration drops the country may suffer a lack of cultural enrichment that makes it diverse and dynamic.
New technology must be implemented in a way that supports humanity, not confuse and destabilize it. We have existed for many years without the “benefit” of AGI and managed to shape the world to suit us. Our “intelligence,” is the intelligence upon which AGI will rest. It has produced a climate crisis, fatal divisions based on race and religion, wars, nuclear weapons, widening wealth inequality, and a soulless economic system driven by financial profit and gain. Our “intelligence” has given us a system where necessities like health, education, food, and shelter have been commodified, available to you if you can afford it. It is true that our intelligence has produced many wonders and benefits. Most of these benefits had to indicate a strong return on investment before they were initiated. The opportunity for financial gain is always a strong motivator. It is seldom for the good of all. We must evolve.
We may have to redefine the idea of working for a living. We may have to rethink the concept of trading money for necessities and goods. We may have to change the concept of profit as we understand it today. We will have to stop reaching for weapons and going to war when we have disagreements with our neighbors. Otherwise, there are going to be many more desperate, angry people to manage and support with the coming of AGI.