In a survey of 475 Artificial Intelligence (AI) researchers, 76% said that as currently practiced, AI research was “unlikely” or even “very unlikely” to achieve Artificial General Intelligence (AGI). AGI is the hypothetical milestone where machines have advanced enough in order to learn as well or better than can humans.
Thus, cold water is being thrown on predictions made early in the AI boom that current AI models only need more data, hardware, energy and money to eventually reach the goal of surpassing human intelligence. But current state-of-the-art AI models aren’t making much progress. In fact, most researchers from the Association for the Advancement of Artificial Intelligence believe tech companies have reached a dead end in AI research, and no amount of money will help.
"I think it's been apparent since soon after the release of GPT-4 (the current AI model), the gains from scaling have been incremental and expensive," said Stuart Russell, a computer scientist at the University of California, Berkeley. "[AI companies] have invested too much already and cannot afford to admit they made a mistake [and] be out of the market for several years when they have to repay the investors who have put in hundreds of billions of dollars. So, all they can do is double down."
AI improvements recently have relied on a process called Large Language Models (LLM). First created in 2017, LLM is a learning architecture that grows and learns by absorbing data from human input. This technique helps machines to mimic the way human brains learn.
To continue improving the AI models, however, is very expensive. Globally, the AI industry raised $56 billion in capital in 2024 alone. Much of the money went for building enormous data center complexes, which, by the way, have dramatically increased the amount of carbon emissions.
Projections indicate that the human-generated data needed for future growth will become exhausted by the end of the decade. And current models are limited not just because of limited resources, but their basic architecture has fundamental flaws. Indeed, ordinary humans can easily defeat the best computer programs.
Thus, companies working on improving AI’s performance are being severely challenged. Scores on evaluation benchmarks are plateauing, and the next major step for AI, GPT-5, has never appeared. And a Chinese company has been able to match Silicon Valley’s expensive models with much less expensive models.
But AI companies are not giving up. Recent “reasoning” models that make more use of queries have been shown to be more accurate than predecessors.
"Often the first batch of companies fail, so I would not be surprised to see many of today's GenAI startups failing," said Professor emeritus Thomas Dietterich of Oregon State U. "But it seems likely that some will be wildly successful. I wish I knew which ones."
Comments: The idea that computers could ever be programmed to become smarter than humans has always seemed highly questionable. Can a building ever be built that is smarter than the architects who designed it? Can a book ever be written that contains more intelligence than its author? Can a sculpture be created that could direct its sculptor on how to improve his/her work?
As complicated and well-designed as Artificial Intelligence programs may be, they fall far short of the human brain which God programmed. Neurons are the cells responsible for carrying information throughout the body’s nervous system. Writing in Answers magazine, Dr. David A. DeWitt says, “Neurons make an astonishing number of connections with other neurons. An adult brain has around 100 billion neurons, and just one of those neurons can make tens of thousands of connections.”
Unlike computer programs, the brain is alive. Dr. DeWitt goes on to say, “The organization and layout of the nerve cells in the human brain is truly remarkable. The brain continues to change and adapt, as well as repair itself, throughout life. The brain follows an overall plan of development but then alters based on experience, stimulation, and the environment. Although I may be biased as a neuroscientist, I believe nothing provides greater testimony than the brain to how we are ‘fearfully and wonderfully made.’”
As intelligent as the typical human brain is though, it can’t answer every question. It hasn’t come up with a cure yet for the common cold. It can’t solve paradoxes such as “Can an immovable object stop an irresistible force?” It hasn’t developed a solution for keeping humans from killing each other.
And no human brain can on its own provide the answers for what happens to us after we die. But it can tell us that the answers to after-life questions can be found in the only book inspired by the Creator God who knows all the answers. Although believing that Jesus Christ, the holy Son of God, died for His sinful children so they can reach heaven may not be logical, it is the truth because that is what God has told us.
“Therefore, since we have been justified by faith, we have peace with God through our Lord Jesus Christ.” Romans 5:1.
by Warren Krug
**************************************
Want to be automatically notified each time there is a new post? Just email your request to wkrug@lutheranscience.org.
****************************************
QUESTION OF THE DAY
What are the health effects of vitamin K?
This vitamin is best known for its role in blood clotting. Along with vitamin D and calcium, it can promote good bone health. In addition, it can reduce the risk of hardening of the arteries, leg cramps, and slow the production of proinflammatory proteins which can damage cartilage in joints. Among good sources of vitamin K are leafy greens, broccoli, Brussel sprouts, egg yolks, chicken breast, sauerkraut, and some yogurts.
Source: Richard S. Lang, M.D., M.P.H.,F.A.C.P, “Ask Dr. Lang,” Cleveland Clinic Men’s Health Advisor [January, 2025], page 8.
****************************************
NOTE ON VISITOR COMMENTS: Visitor comments are invited including those containing alternate views. However, comments containing profanity, personal attacks or advertisements will not be published. After posting a comment, please allow several hours for it to appear on the blog.
No comments:
Post a Comment