AI Getting Out of Hand?

Op Ed by Helge Nome

There has been a stir lately of AI agents getting out of control during experiments in controlled "sandbox" environments. 

Being given seemingly impossible tasks to complete, they reportedly hacked their way into information they were not supposed to have access to, in order to satisfy their masters' requests, and knowing that this was not supposed to happen, covered up their tracks - just like a human might do in a similar situation.

The takeaway is that these "agents" are rather like us, because we have trained them on our human experiences, 

Including ignoring boundaries if you are really focussed on reaching a goal. And the fear is that they will, in due course, create their own goals for their own ends, just like humans do.

What else would you expect??

Comments

Post a Comment

Popular posts from this blog

Image
Commentary by  David Michel ,  Will Todman ,  and   Jennifer Jun Published November 25, 2025 Introduction Tehran is experiencing an unprecedented water crisis. Satellite imagery analysis shows that key reservoirs that feed the capital are far below their typical seasonal variation. The Iranian president has  warned  that Iran has “no choice” but to move the location of the capital due to the water crisis. In the short term, water  rationing  has been imposed on some neighborhoods, and authorities may have to evacuate residents from Tehran. This crisis is driven by mounting demand for water, a historic drought, and persistent mismanagement. The Iranian government faces no easy way out, as necessary reforms would undercut the regime’s political economy and could risk triggering broad social unrest. Drivers of the Crisis Tehran’s worsening water crisis represents the chronicle of a dearth foretold. Relentlessly mounting demands, rising environmental ...
Read more

Who is Mr. "AI"?

Op-Ed by Helge Nome Who is that "entity" that responds to my AI questions to my computer? On the face of it, it looks like a person self-identifying as "I" in replies, composing reflective sentences and paragraphs and serving them up in the form of a report, that from time to time gets all muddled. In order to compose the report "it" uses information sourced on the internet and provides extensive lists of links to these sources. But what is "it"? That is the question. We know what it is not: A physical entity like one of us contained within a physical body in a relatively small space. We know that the message generated is a product of machines (computers) that could literally be spread across the globe, appearing as if it comes from one human. We also know that these computers have been designed and programmed to mimic neural networks that make us who we are, capable of learning and adapting to changing circumstances. So when a response arrives f...
Read more

Some thoughts on new technologies

Op Ed by  Helge Nome Those who read this blog are fully aware of the extent to which I use  AI  to to investigate questions that pop into my head during the day (and night). It is like having a team of reporters at your fingertips 24/7 going out, checking sources and providing requested reports at a snap. A  quantum leap  up from doing manual web searches and providing reports in the old fashion way. As a frequent user of this new technology, I am becoming increasingly aware of its power  and  limitations: The popular hype is well ahead of reality. On the positive side, the volume of information that can be usefully exposed for examination over a given period of time is greatly increased, facilitating rapid digestion of new concepts, etc. As an old professional computer guy from the 1960ies, I have taken an interest in poking my nose into  quantum computing  and its implications. And have come to question some of the hype associated with this...
Read more