It is not unimaginable for “bad actors” to develop malicious AI. And it is certain that AI will alter the wealth of nations, and impact the vulnerability of economies. Margrethe Vestager the technology chief of the EU is already looking at a voluntary code of conduct for AI.
A central challenge of AI is alignment. How do we build an AI system that is aligned by design with the goals of humanity? What underpins human values? Is it possible to isolate values that are shared by all peoples across all cultures? Is it possible to represent them in such a way that an AI system can understand and take them into account in making decisions?
Once an AI system achieves a particular quality of intelligence, it will design better AI than humans can. It will improve its own capabilities or build other AI systems that are more intelligent than itself.
Researchers now believe that humanity must look deeper at AI bias, predictive policing, and misinformation. It is already the case that deepfaking is blurring fact and fiction. Deepfakes depend on the use of deep neural networks that employ a face-swapping technique. Emerging synthetic media has been boosted by of a slew of new “generative AI” tools like Midjourney.
AI tools are now ubiquitous. But more than that is the fact that AI has now abruptly closed the Digital Divide between those who are programmers and those who are not. AI has made everyone a programmer. The programming barrier is now incredibly low. AI has initiated a New Programming Era. The efflorescence of this AI-driven change will vine rapidly across every industry.
It is a dramatic shift and it has spawned the need to regulate the emergence of Superintelligent AI into society. Over the next decade, we will witness AI systems that will exceed expert skill levels in most domains, delivering productive activity resembling one of today’s largest corporations.
Hugging Face, Inc., uses Open-Source and Open-Science to build tools using machine learning. Its transformers library, built for natural language processing applications, permits AI aficionados to share machine learning mock-ups and datasets.
Ilya Sutskever, the chief scientist at Open AI, has called for a monitoring agency that can place restrictions on degrees of AI deployment, levels of security, testing for compliance, routing audits, and systems inspections.
We continue to see behind the next obstacle, and where we once used Stone Age Chips, today we deploy Microchips. We continue to recraft a world we make with our minds. Today, tools, like algorithms, have the ability to take actions to achieve predetermined goals, and others linked to AI have the ability to understand human speech.
Using a microchip made using AI, a paralysed man walked by just thinking about it. Using AI, researchers wishing to recraft the world exposed a bacteria called acinetobacter baumannii to thousands of drugs where the precise chemical structure of each compound was known. The AI was then unleashed on 6,680 compounds whose effectiveness is unknown.
In ninety minutes, AI produced a shortlist. Two hundred and forty were further tested and nine were identified as potential antibiotics. “Abaucin” emerged as an extremely powerful antibiotic. Many antibiotics kill a spectrum of bacteria. Curiously, this antibiotic works only on acinetobacter baumannii. The exactness of “abaucin” makes it difficult for drug resistance to develop. And this may also result in fewer side effects during treatment.
Antibiotics are used to kill bacteria. But a paucity of new antibiotics for decades means that bacteria are becoming harder to treat, as bacteria evolve resistance to known antibiotics. Over a million people die annually from infections that resist treatment with antibiotics.
Acinetobacter baumannii is one of the most problematic species of bacteria. The WHO has identified it as a “critical” threat. It is often resistant to multiple antibiotics and is a problem in clinics and hospitals, where it survives on surfaces and medical equipment.
In New York, in May 2023, a firm used ChatGPT for legal research. After filing pleadings, the judge met the unprecedented circumstance of reference example legal cases which exist nowhere. The firm was unaware that the content generated using AI tools could be false or inaccurate. Six of the cases used in the submissions are not available on any known legal reference database such as LexisNexis or Westlaw.
The biomedical scholar Robert Arp describes imagination as “scenario visualization”. He traces this to our hominid history that once allowed us to re-envision a stick to become a stake, and by further imaginative leaps – a hurling spear. Anticipation allows us to see behind the next obstacle; to see what is not there, at least – not yet.
Anticipation allows us to recraft the world. It allows us to act to realize a world that we re-imagine constantly. Misunderstandings, memory, and anticipation intervene creatively with our thoughts, not solely to represent or record our experiences, but to exceed and even distort them.
The effect of imagination in this process is the fertilizing of ideas into prolific results. This way, we are able to envisage outcomes ahead of their appearance. Endings that are invisible, impossible, and unexampled in experience.
