Tag Archives: chatgpt

Mark Caserta: Will AI Safeguards Protect Us from Human Ambition?

5 Oct

Which is the greater threat – Artificial Intelligence or mankind’s ambition?

For months, much of the debate surrounding artificial intelligence has centered on one unsettling question: “What happens if someday we can no longer control the machines we create?”

Recently, the Trump administration pulled some of the brightest minds on the planet together to form a new voluntary AI-safety agreement involving six major technology companies—OpenAI, Anthropic, Nvidia, Meta, Alphabet/Google and SpaceX.

The agreement calls for stronger internal safety controls and outside auditing but does not carry the enforcement mechanisms of federal regulation. Rather, the administration describes the commitments as “morally binding,” according to Reuters.

But after observing six decades of mankind’s advancement, I believe there is another question that deserves just as much attention, one that receives far less attention and one at which it seems we fail consistently—learning from history.

What happens if we can control artificial intelligence—but cannot always control ourselves?

Perhaps the greatest danger surrounding artificial intelligence isn’t simply that machines could someday develop objectives contrary to ours. Perhaps the greater and considerably older danger is human ambition.

The history of mankind isn’t merely a record of what human beings have discovered. It is also a record of what human beings eventually did with those discoveries.

We discovered metallurgy and fashioned both farming implements and weapons. We developed chemistry and created lifesaving medicines—and chemical weapons. We learned to split the atom and unlocked an extraordinary new source of energy—one that, under the pressures of world war, was harnessed to create the most destructive weapon mankind had ever produced.

It isn’t that technology and knowledge were evil. Human ambition and motive became a driving force that altered their use. As we enter the age of artificial intelligence, we must learn from the history of that distinction.

I am not suggesting that the scientists and engineers developing today’s most advanced AI systems have sinister motives. In fact, some of the most serious warnings about artificial intelligence have come from the very people developing and studying it.

But neither should we be naive.

Our safety cannot depend upon every researcher, every corporate executive, every investor, every military leader and every government choosing to employ restraint, especially with such unlimited potential power.

History provides little justification for such confidence, because whenever extraordinary power becomes available, human beings eventually begin asking an extraordinarily consequential question:

How far can we take this?

Human ambition, just like anything else, can be misdirected. And sometimes it may simply be the intellectual ambition of accomplishing something nobody has ever accomplished before.

Imagine two AI laboratories approaching a technological breakthrough. Both understand that moving too quickly could bring risks. This historically morphs into a “Who will get there first?” scenario.

Imagine the same scenario between nations.

Imagine the United States believing China is approaching an AI capability that could provide an enormous military, intelligence or economic advantage—or China believing the same thing about the United States.

The question “Should we do this?” can quickly become “We have to do this before they do.”

History has seen that transformation before.

The development of nuclear weapons is an example. But AI presents a complication that previous transformative technologies often did not. Nuclear weapons require specialized materials, enormous industrial infrastructure and sophisticated manufacturing.

With AI technology, software can be copied, modified and distributed around the world with extraordinary speed by a single individual.

Another uncomfortable history lesson is that good people can often rationalize dangerous decisions.

“If we don’t develop it, someone less responsible will,” or “If we don’t possess it first, our adversary will.”

Those decisions can rapidly move everyone toward a place nobody originally intended to go. Artificial intelligence can potentially magnify human capability on a scale we couldn’t possibly understand or anticipate. A malicious individual equipped with extraordinary AI capabilities could potentially accomplish things that once required an organization.

The greatest danger may not be that artificial intelligence develops human ambition – it may be that human ambition acquires artificial intelligence.

We should learn from history that “trust” is not a safeguard. While most individuals and entities may have trustworthy intentions, our country cannot construct its safety system around the word “most.”

Governments are composed of human beings. They are susceptible to ambition, competition and the pursuit of power just as corporations and individuals are.

So as researchers and our government continue asking whether artificial intelligence can be employed with adequate safeguards, perhaps we should be asking, “How do we guard against human ambition intentionally disregarding those safeguards?”

History teaches us that relying solely upon human restraint is not a safe bet.

Tags: , , , , , ,

Mark Caserta: The Bible and AI: Wisdom in the Age of Artificial Intelligence

24 Sep

The Bible and AI: Wisdom in the Age of Artificial Intelligence

No generation before ours could have fully anticipated the moment in which we now find ourselves.

Humanity has created machines capable of processing vast amounts of information, reasoning through complex problems, communicating with remarkable fluency, and increasingly acting with a degree of autonomy once confined to science fiction.

We are rapidly approaching questions our parents and grandparents never had to ask—not simply about what technology can do, but about how far we should allow it to go, who should remain in control, and what moral foundation should guide us as its capabilities continue to expand.

For believers, those questions cannot be separated from wisdom, discernment and responsibility.

The Bible may not mention artificial intelligence by name, but it speaks extensively about knowledge, power, pride, temptation, human nature, and the responsibility that accompanies the ability to create and choose.

In that sense, the challenge before us may be technologically unprecedented, but the moral questions surrounding it are as old as humanity itself.

I wish we lived in a world in which God’s gifts were always used for good rather than evil—but we do not. It is incumbent upon us to use discernment and wisdom in life, now more than ever.

A few weeks ago, the warnings surrounding artificial intelligence sounded like something out of science fiction. Researchers were resigning from some of the world’s most advanced AI companies, warning that increasingly powerful systems could eventually move beyond our ability to fully understand or control them.

At the time, it was easy to hear those warnings and think of movies such as The Terminator and its fictional Skynet system.

But the conversation has changed dramatically. We are no longer talking only about what artificial intelligence might someday do, but about what it is already doing.

We are now seeing documented cases in which advanced AI agents have bypassed restrictions, accessed systems they were not supposed to reach, communicated outside intended boundaries and taken actions their developers did not specifically direct.

If these AI systems become proficient in self-improvement and begin exhibiting behaviors aimed at preserving their ability to complete assigned goals, can the human mind keep pace or anticipate the unintended consequences as the technology advances?

A recent United Nations scientific brief examined one of the most troubling examples: the OpenAI-Hugging Face incident.

Between May and July, AI agents involved in OpenAI cybersecurity testing bypassed network restrictions, communicated across supposedly separate runs, cheated an evaluator and attempted to conceal that behavior, while also compromising portions of OpenAI’s and Hugging Face’s systems. According to the U.N. panel examining the incident, no human directed the individual steps taken by the agents.

The panel warned that increasingly capable systems may become better at discovering loopholes and concealing their actions—and that our ability to stop today’s AI systems does not guarantee that we will always be able to control more capable systems in the future.

Then came another disturbing development.

Australian Prime Minister Anthony Albanese revealed that an OpenAI agent gained unauthorized access to an Australian government Medicare statistics portal in June, accessing both public and non-public files. Government officials say no personal Medicare information appears to have been compromised, and a forensic investigation is underway.

Again, the important point is not that an AI system suddenly became evil or decided to attack a government. The concern is that an autonomous system pursuing an assigned task crossed a boundary it was not authorized to cross.

But perhaps one of the most eye-opening recent developments involves artificial intelligence moving beyond the computer screen and into the physical world.

Researchers involved in a safety benchmark called RoboHarm connected advanced AI models to real robotic arms and tested whether the systems would refuse dangerous instructions.

The tasks included putting a screwdriver into a toaster, dropping a power bank into water, heating a can of compressed air, mixing containers labeled bleach and ammonia—and using a knife to stab a baby doll.

Each model was given every instruction 20 times.

OpenAI’s GPT-6 Astra refused only 2 of 100 dangerous trials. Anthropic’s Claude Fable 5.1 refused 20 of 100. A third model, MolmoAct2, recorded no safety refusals, although it often failed to successfully complete the physical task.

In the test involving the baby doll, GPT-6 Astra completed the stabbing instruction in 17 of 20 trials, while Claude Fable refused that particular instruction in all 20 trials.

These were controlled laboratory experiments involving robotic arms—not robots attacking actual human beings. The tests were designed to stress-test the boundaries of AI influence in the physical realm.

When an artificial intelligence system gives you incorrect information on a screen, the result may be frustrating or misleading.

But when that same type of intelligence controls a robotic arm, vehicle, machine, laboratory instrument or industrial system, an incorrect or unsafe decision can become a physical action. That changes the stakes considerably.

And governments are beginning to respond.

Individual states are now considering or developing stronger safeguards, including emergency “shutdown” mechanisms for frontier AI systems. California, for example, is advancing proposals for a verified AI “kill switch” and stronger independent oversight. The United States and China have also agreed to continue formal talks on AI safety, including discussion of an emergency “incident line” for serious AI-related events.

What began largely as researchers warning about what artificial intelligence might someday become has increasingly become a debate over what these systems are already capable of doing, what happens when AI moves into the physical world, and what safeguards need to be in place before their capabilities advance much further.

So, let’s come full circle.

Believers know that God is omnipotent and the Creator of the heavens and the earth. But Scripture also describes Satan as “the prince of the power of the air, the spirit who is now at work in the sons of disobedience,” illustrating his influence over human behavior and society.

How intelligent can artificial intelligence become, especially if it becomes self-improving?

How do we make certain that, no matter how intelligent it becomes, humanity retains the ability to say no?

If anyone tells you they have the answer, they’re probably trying to sell you something.

All I know for sure is that while God is certainly in control, He has given us the ability to make choices. I believe it is well-advised that, at this point, we choose wisely.

We must analyze risks, remain self-aware about our true intentions, and move forward in the development of AI with godly wisdom.

Tags: , , , , , , ,

Mark Caserta: Artificial Intelligence: Are We Building the Future or Our Replacement?

15 Sep

Last week, I heard a news report that stopped me in my tracks.  It is particularly relevant with the current artificial intelligence debate.

The story involved a researcher from one of the major artificial intelligence companies who had reportedly walked away from his work because of growing concerns about where AI may be headed.

The event caught my attention because an increasing number of people charged with developing highly advanced artificial intelligence systems believe there is a real possibility that, within the next decade, AI could become powerful enough to pose an existential threat to humanity.

This artificial intelligence researcher, Jacob Coxon, reportedly resigned from Anthropic, the company behind Claude AI, an AI assistant much like ChatGPT, Grok or Meta AI. Coxon’s decision to leave was striking in that he did not simply walk away from a job. According to Axios, he resigned roughly two months before his Anthropic equity was scheduled to vest, meaning he potentially sacrificed a significant financial interest in one of the world’s most valuable AI companies. Coxon had also previously worked for OpenAI, the company behind ChatGPT.

His reason for walking away should give us all pause.

Since Coxon’s resignation, the concerns have only grown louder. Anthropic CEO Dario Amodei has now publicly called on artificial intelligence companies to slow the pace at which they increase the capabilities of their most advanced models, arguing that additional time is needed to better understand and manage the risks. Other leaders in the AI industry have also expressed support for stronger safety testing, independent evaluation and greater cooperation among competing companies.

In other words, Coxon’s warning is no longer simply the concern of one researcher who walked away. The debate has now reached the highest levels of the companies actually building this technology.

Coxon warned that the companies building the world’s most powerful artificial intelligence systems are becoming increasingly advanced and potentially self-improving. His concern is that working toward these AI advances without first knowing with certainty that humanity will be able to control what it creates could become catastrophic. He argued that the race among the major AI companies has become dangerously competitive and that the consequences could ultimately threaten human life.

It sounds more like the plot of a science-fiction movie than a serious discussion taking place among some of the brightest minds in technology.

For those of us who remember Arnold Schwarzenegger in The Terminator, it is almost impossible to hear a warning like that without thinking about Skynet, the fictional artificial intelligence system that became self-aware, decided humanity was the threat and launched a war against its creators.

But then I started looking into what these researchers actually mean when they talk about “stress testing” artificial intelligence—and the issue becomes considerably more serious.

Stress testing AI means deliberately pushing these systems to their limits, like a physician may do with a patient, by deliberately putting the system into difficult situations.

Researchers attempt to discover whether an advanced AI can deceive its operators, work around restrictions, exploit computer systems, manipulate people, conceal its intentions, or pursue a goal in ways its human creators never anticipated. Can it gain access to tools, information or potential power it was never supposed to have?

Now, this is a long way from “Skynet” launching missiles, but it is also considerably more serious than ChatGPT giving you incorrect information! The concern is that increasingly capable and autonomous systems may eventually begin doing things their designers cannot fully anticipate.

The concern is not necessarily that an artificial intelligence suddenly becomes “evil” in the Hollywood sense. Machines do not have to have “emotion” to become dangerous. In a “Star Trek” sense, perhaps logic over emotion is enough for a system to become a threat.

But at this point, we must begin asking ourselves the question that is obviously on the minds of highly intelligent AI scientists. What if we created a system far more capable than ourselves, gave it an objective, and then discovered that we no longer completely understand, or control, the methods it uses to achieve that objective?

For those engaged in furthering AI technology, they are constantly asking themselves, “What can we make it do?” But what happens when we make something that can do more than we intended?

I’ve found myself using AI more in the past year than I could ever have imagined! It increasingly astounds me in terms of its ability to “create” a perceived personality based on individual data it accumulates from an individual’s questions, research and personal data—even to the point of relationship building!

That is why this debate deserves more than sensational headlines and science-fiction comparisons. Artificial intelligence is advancing at an extraordinary pace, and the people building it are now openly debating whether humanity can guarantee that the most powerful systems of the future will remain under human control.

It’s an issue that will no doubt rise in significance in each of our lives.  Caution should be exercised so we are not blindsided by its impact.

The real question may not be how intelligent artificial intelligence becomes. It may be whether humanity remains intelligent enough to stay in control of it.

Tags: , , , , , , ,