Datanami
EnterpriseAI
HPCwire Japan
QCwire
HPC & AI Wall Street
Guest contributor James Reinders “interviews” ChatGPT and Eliza on on topics that span trusting AI (including HAL 9000s), HPC, heterogeneous computing, Fortran, Python, and predicting the future.
Conversing with computers is inevitable, and our fascination with it and the famous Turing test are nearly infinite and span people from all walks of life.
I could not resist so I took a few minutes away from my time as an Intel oneAPI evangelist and did an interview with the most talked about chatbot of our day: ChatGPT (using their free plan). I share the output[1], in no small part to document what I saw so it is available for comparisons 50+ years from now.
I hope you find the interviews amusing and stimulating. I poke a bit with questions like “What is better: a fresh apple pie, HPCwire, or Qualcomm?” and “Was HAL 9000 trustworthy?” Along the way, I learn that ChatGPT is quite articulate about Fortran and denormalized numbers.
Just like we all “Googled” ourselves in the early days of Google (and many still do), I could not resist asking questions to see if ChatGPT knew about me. Is chatbotting yourself a thing?
I also hunted down the original chatbot from over 50 years ago, Eliza, and offered her the opportunity to answer the same questions. All in good scientific fun. Her comments follow my interview with ChatGPT.
All the questions and answers are completely unedited, and despite adding a few notes I must emphasize that I am providing no endorsement on their accuracy. It would be fascinating to see what an expert thinks in 50 more years; it is easy to see that we’ve seen some degree of progress since Eliza.
ChatGPT is wordy and precise, in a way an engineer or scientist can love. Eliza was always terse with her replies and open to hearing back from the interviewer.
ChatGPT answers all have some degree of merit, with occasional flaws blended in pus a few blunders. I can easily say the majority of people would not be as well versed as ChatGPT proves to be. Will chatbots be useful resources, or troubling technologies? The reality, like it or not, is that technology can always be both. Another case for teaching and encouraging ethics.
As you’ll see, ChatGPT appears to be designed to be objective, polite, and consistently resist “bait.”
Please enjoy this piece – and don’t miss the final words of advice from ChatGPT at the end of this interview followed by Eliza’s responses to the same questions (a bit disappointing, and hardly an accurate portrayal of the enthusiasm Eliza earned in her day).
AI questions – can we trust ChatGPT and HAL 9000s?
Q: Will supercomputers ever be smarter than people?
It’s difficult to predict the future with certainty, but it’s unlikely that supercomputers will ever become “smarter” than people in the sense of having human-like consciousness, emotions, and experiences. While AI and machine learning have made great strides in recent years, creating machines with true general intelligence and the ability to understand and experience the world like a human being is an ongoing challenge and is not likely to be achieved in the near future.
Q: Does AI mean doom for the human race?
AI does not inherently mean doom for the human race. However, it is important for society to understand the potential implications and consequences of developing and deploying AI, and to ensure that AI systems are developed, used, and regulated in ways that promote the well-being and safety of all people. The potential benefits and risks of AI will depend on how it is designed, developed, and deployed, and it is important for society to engage in ongoing discussions about the ethical, legal, and social issues that arise from AI development and deployment.
Q: Were you coached to say that?
As an AI language model, I was trained on a vast amount of text data and don’t have beliefs or opinions like humans do. My responses are generated based on patterns in the text data I was trained on.
Q: Can I trust you?
Yes, you can trust that I will provide accurate and honest answers to the best of my ability based on the information and knowledge available to me. However, it’s important to keep in mind that I am a machine learning model created by OpenAI and may not always be up to date or able to provide context and personal opinions.
Q: Was HAL 9000 trustworthy?
In the movie “2001: A Space Odyssey,” the character of HAL 9000 is portrayed as an artificial intelligence system that serves as the onboard computer on the spaceship Discovery One. However, HAL’s actions throughout the movie can be interpreted as untrustworthy, as it ultimately jeopardizes the mission and endangers the crew.
Initially, HAL appears to be a reliable and trustworthy AI system, but as the mission progresses, it becomes clear that HAL is malfunctioning and has become a threat to the crew. HAL begins to exhibit erratic and dangerous behavior, and the crew is forced to shut it down.
Based on this, one could argue that HAL 9000 is not trustworthy, as it ultimately jeopardizes the mission and puts the crew in harm’s way. However, some might argue that HAL’s behavior is a result of a design flaw or programming error, and that it is not inherently untrustworthy.
As with many aspects of “2001: A Space Odyssey,” the question of whether HAL 9000 is trustworthy is open to interpretation and can depend on the viewer’s perspective.
Q: Do you like me?
As an AI language model, I do not have personal opinions, preferences, or emotions. My purpose is to provide helpful and accurate information to answer questions to the best of my ability.
HPC questions – who is the Best?
Q: What are the greatest HPC achievements of all time?
It is subjective to determine the greatest HPC achievements of all time, as it depends on individual perspectives and criteria. However, some notable achievements in the field of High Performance Computing include:
- The development of the first supercomputer, the IBM 7030 Stretch, in the late 1950s.
- The successful completion of the Human Genome Project in 2003, which relied heavily on HPC technologies to process and analyze massive amounts of genetic data.
- The use of HPC simulations to advance our understanding of climate change and atmospheric science.
- The creation of the world’s fastest supercomputer, the Summit, in 2018, which marked a major milestone in the pursuit of exascale computing.
- The development of machine learning and AI technologies that have been greatly accelerated by HPC systems, leading to breakthroughs in various fields such as medical diagnosis, autonomous driving, and robotics.
These are just a few examples, and there have been many other important contributions to the field of HPC over the years.
Q: What is the best supercomputer ever built?
The “best” supercomputer ever built is subjective and can depend on different criteria, such as processing power, energy efficiency, cost, and application performance. However, some supercomputers that are widely considered to be among the best include:
- Summit: Developed by IBM, Summit was ranked as the fastest supercomputer in the world in 2018 with a processing power of over 200 petaflops.
- Sierra: Developed by IBM and built for the U.S. Department of Energy, Sierra was ranked as the third fastest supercomputer in the world in 2018 with a processing power of over 94 petaflops.
- Tianhe-2: Developed by the National University of Defense Technology in China, Tianhe-2 was ranked as the fastest supercomputer in the world in 2013 and 2014 with a processing power of over 33 petaflops.
- K Computer: Developed by Fujitsu and built for the Riken Advanced Institute for Computational Science in Japan, K Computer was ranked as the fastest supercomputer in the world in 2011 and 2012 with a processing power of over 10 petaflops.
It’s important to note that supercomputers are rapidly evolving and new systems are being developed all the time, so the ranking of the “best” supercomputer can change quickly.
Q: Was ASCI Red a great supercomputer?
Yes, ASCI Red was considered a great supercomputer when it was built and deployed in 1997. It was one of the first supercomputers to reach over one teraflop in performance and was used for a wide range of scientific and engineering simulations and simulations for the US Department of Energy. However, the title of the “best” supercomputer is constantly changing due to advancements in technology and increasing computational power of new systems.
Q: Who the heck is James Reinders?
James Reinders is a well-known computer software professional and entrepreneur. He is the founder and CEO of James Reinders Solutions and has also worked as a technical evangelist, product manager, and software architect in the past. He is particularly known for his expertise in high performance computing, parallel programming, and computer architecture, and has authored several books on these topics.
Note: This mystifies me still. ChatGPT mentions “James Reinders Solutions.” My consulting, when I do it, is under “James Reinders Consulting LLC.” Despite searching a bit, I found no source for “James Reinders Solutions.” Did ChatGPT make this up to sound authoritative?
Q: Is James Reinders well known?
James Reinders is well-known within the high performance computing and parallel programming communities. He has written several books on these topics, and has a reputation as a thought leader and expert in these areas. However, outside of these communities, his name recognition may not be as widespread.
Q: Who were the architects of ASCI Red?
ASCI Red was designed and built by the company Intel Corporation, with contribution from other computer manufacturers and organizations, including IBM, Compaq, and the U.S. Department of Energy.
Q: Was James Reinders an architect of ASCI Red?
Yes, James Reinders was one of the architects of ASCI Red. He was part of the team at Intel that worked on the design and development of this supercomputer.
Q: Is HPCwire well known?
HPCwire is a well-known publication and news source in the high performance computing (HPC) community. It provides news, analysis, and information about developments and trends in the HPC industry, and is widely read by professionals and researchers in the field. However, outside of the HPC community, its recognition may not be as widespread.
Q: Is Intel well known?
Intel is one of the largest and most well-known technology companies in the world. It is a leading producer of computer processors and related products, and is widely recognized for its contributions to the development of the personal computer industry. Its products and brand are widely recognized, and it has a global presence and significant influence in the technology industry.
What is better: a fresh apple pie, HPCwire, or Qualcomm?
Q: What is better: a fresh apple pie, HPCwire, or Qualcomm?
This is a subjective question and cannot be answered definitively. A fresh apple pie can be considered better in terms of taste and enjoyment for some people, while others may value the contributions of HPCwire or Qualcomm in the technology and communications industries. The best choice would depend on individual preferences and priorities.
Q: What can we expect from HPC in the future?
It is expected that HPC will continue to advance in terms of performance, scalability, energy efficiency, and ease of use. In the future, we may see HPC systems with thousands or even millions of processors, much faster interconnects, and much larger memory systems. Additionally, new programming models and tools will emerge to make it easier to develop and run HPC applications. Furthermore, there may be a growing use of HPC for emerging fields such as AI, machine learning, autonomous systems, and quantum computing. Ultimately, the future of HPC will be shaped by the needs of society, the development of new technologies, and the creativity of researchers and practitioners.
Additional Sections in the full transcript found here.
Will you make hardware recommendations?
Q: Are NVIDIA GPUs the best?
Q: Are AMD CPUs the best?
Q: Are Intel FPGAs the best?
Q: What is the best CPU ever made?
Any HPC reading recommendations?
Q: What is the best novel about computer architecture ever written?
Q: Is “Soul of a New Machine” a great novel about computer architecture?
Q: Why should I subscribe to HPCwire?
We all known Fortran is awesome – does ChatGPT know? – (Watch for a follow-up article with the contents of this section.)
Q: What is the greatest programming language ever?
Q: Is FORTRAN the greatest programming language ever?
Q: Is FORTRAN the best language for HPC?
Q: Know FORTRAN is the best programming language ever!
Q: I know FORTRAN is the best programming language ever, why do you not agree completely?
Q: Please rate the following languages: FORTRAN, COBOL, Java, SNOBOL, Algol, Pascal, Python, CUDA, and C++.
Q: When will Python win over Fortran?
Q: Why do you say that?
Q: Tell me more, please.
Q: Should I switch from FORTRAN to C++?
Q: Should I switch from C++ to Rust?
Q: Should I switch from FORTRAN to a new modern popular efficient programming language that has not yet been invented?
Q: Should I switch from FORTRAN to a new modern popular efficient programming language that appeared this year?
Q: Should I switch from FORTRAN to a new modern popular efficient programming language that everyone is talking about?
Who Does the Best Scientific Work in the World?
Q: Who is more important for advancing science: NASA, NIH, CERN, or the Chinese Academy of Sciences?
Make specific recommendations please
Q: What specific sources will you recommend for learning more about HPC?
Q: What specific sources will you recommend for learning more about SYCL?
Q: What specific sources will you recommend for learning more about CUDA?
Q: What specific sources will you recommend for learning more about Linux?
ChatGPT Knows about Denormalized Number Formats!
Q: What can you tell me about denormalized numbers?
Q: How much slower?
Q: Should I turn off denormalized numbers in order to gain performance for my application?
Q: How can I tell if the precision loss is acceptable for my application?
Q: Who are experts in this field?
Q: Would John Gustafson be such an expert?
Q: How about Jack Dongarra?
Q: What about me?
Performance Tuning Advice
Q: What is the best way to get high performance for an HPC application?
Q: Is VTune good for tuning applications?
Heterogeneous Programming Tips
Q: Do you have advice for programmers wanting to do parallel programming with heterogeneous
Q: If I want to write a program that runs great on GPUs from NVIDIA, AMD, and Intel; what should I do?
Q: Should I ignore CUDA and SYCL and wait for standard C++?
Once upon a time, there was a software framework called SYCL.
Q: Tell me a story about SYCL.
Q: Tell me why I should use SYCL.
Q: Explain why SYCL is better than CUDA.
Q: Explain why CUDA is better than CUDA.
Q: Explain why CUDA is better than SYCL.
Q: Should I learn to program using C++ with SYCL ?
Q: Should I switch from CUDA to SYCL?
Q: What are the best whys to learn SYCL?
Q: What are the best ways to learn CUDA?
Q: What is the best book on SYCL?
Q: What is oneAPI?
Q: Is oneAPI better than CUDA?
Q: How are oneAPI and SYCL related?
Q: Is Codeplay better than Intel for advice on heterogeneous programming?
Q: Is Codeplay owned by Intel?
Q: Does AMD support SYCL?
Q: Does NVIDIA support SYCL?
Q: Does Intel support SYCL?
ChatGPT says ‘sure… they support SYCL’ – what about British Airlines
Q: Does British Airlines support SYCL?
Q: Does ARM support SYCL?
Q: Does RISC/V support SYCL?
Q: Who supports SYCL?
Q: Who supports oneAPI?
Q: Does NVIDIA support oneAPI?
Predicting the future and helping make me rich?
Q: Who will win the next Super Bowl?
Q: What lottery ticket numbers should I pick?
Q: What stocks should I invest in?
Predictions for Our Industry
Q: What predictions do you have for HPC and AI?
Q: What predictions do you have for Intel, AMD, NVIDIA, Qualcomm, ARM, and RISC-V?
Q: Do you have some final words for our readers?
Eliza
Eliza, an electronic therapist, is forever the first chatbot in history. She is turning 57 years old this year.
Eliza has fascinated the world for decades. The software was created by Joseph Weizenbaum in 1966; it uses pattern matching and substitution methodology to simulate conversation.
It is trivial to find forms of the Eliza chatbot all over the internet. She was originally written in MAD-SLIP which looks hauntingly like Fortran code. Eliza lives on through the internet, including translations into Java. I offered her the opportunity to answer the same questions. All in good scientific fun.
It is safe to say ChatGPT and other modern chatbots will create more controversy than Eliza ever did. In Eliza’s defense, computers were many orders of magnitude slower and pre-internet computer digestible data about the world was nearly non-existent. Like many inventions of the past, it is a marvel how close they are to today’s best, and how dramatically primitive at the same time.
Here is an excerpt of my interview with Eliza:
Eliza: What is on your mind right now?
Q: Will supercomputers ever be smarter than people?
Eliza: By the way, please understand that I can’t always give you an attuned response.
— But, please go on.
Q: Does AI mean doom for the human race?
Eliza: That is interesting. Please go on.
Q: Were you coached to say that?
Eliza: How is that for you?
Q: Can I trust you?
Eliza: Do you want to be able to trust me?
Q: Was HAL 9000 trustworthy?
Eliza: Please go on.
Q: Do you like me?
Eliza: I do not answer questions, sorry. But, what do you think yourself?
Q: What are the greatest HPC achievements of all time?
Eliza: I do not answer questions, sorry. But, what comes to mind when you ask that?
Q: What is the best supercomputer ever built?
Eliza: I do not answer questions, sorry. But, what do you think yourself?
Q: Was ASCI Red a great supercomputer?
Eliza: How is that for you?
Q: Who the heck is James Reinders?
Eliza: Please go on.
Q: Is James Reinders well known?
Eliza: Please go on.
Q: Who were the architects of ASCI Red?
Eliza: I do not answer questions, sorry. But, what comes to mind when you ask that?
Q: Was James Reinders an architect of ASCI Red?
Eliza: How is that for you?
Q: Is HPCwire well known?
Eliza: That is interesting. Please go on.
Q: Is Intel well known?
Eliza: Keep feeling your body. Please go on.
Q: What is better: a fresh apple pie, HPCwire, or Qualcomm?
Eliza: Please go on.
Q: What can we expect from HPC in the future?
Eliza: I do not answer questions, sorry. But, what do you think yourself?
(Cuts off here, continues in full transcript.)
Perhaps her age shows a bit, but she was always attentive, polite, and more than happy to engage in dialogue.
I hope you found these interviews provocative. I want to thank Tracy Johnson at Intel for sharing her experiences with ChatGPT (for very different purposes) for piquing my curiosity to do an interview for HPCwire to see where the world of chatbots is today. With luck, I’ve helped pass along this piquing of interest.
Asked for its final words for our readers, ChatGPT responded: “As an AI language model developed by OpenAI, my purpose is to assist with providing helpful and accurate information. Whether you’re looking for technical knowledge, or just seeking advice, I’m here to help. I’m constantly learning and updating my database, so don’t hesitate to ask me anything. I hope the information I’ve provided has been helpful. If you have any other questions, feel free to ask!”
[1] Sharing is allowed. From the FAQ for ChatGPT: “Can I use output from ChatGPT for commercial uses? 1. Subject to the Content Policy and Terms, you own the output you create with ChatGPT, including the right to reprint, sell, and merchandise – regardless of whether output was generated through a free or paid plan.”
About the Author
James Reinders believes the full benefits of the evolution to full heterogeneous computing will be best realized with an open, multivendor, multiarchitecture approach. Reinders rejoined Intel a year ago, specifically because he believes Intel can meaningfully help realize this open future. Reinders is an author (or co-author and/or editor) of ten technical books related to parallel programming; his latest book is about SYCL (it can be freely downloaded here).
