Visitor contributor James Reinders “interviews” ChatGPT and Eliza on on matters that span trusting AI (together with HAL 9000s), HPC, heterogeneous computing, Fortran, Python, and predicting the longer term.
Conversing with computer systems is inevitable, and our fascination with it and the well-known Turing take a look at are just about limitless and span other people from all walks of lifestyles.
I may now not withstand so I took a couple of mins clear of my time as an Intel oneAPI evangelist and did an interview with probably the most mentioned chatbot of our day: ChatGPT (the use of their loose plan). I percentage the output[1], in no small section to report what I noticed so it’s to be had for comparisons 50+ years from now.
I am hoping you to find the interviews a laugh and stimulating. I poke a bit of with questions like “What is healthier: a contemporary apple pie, HPCwire, or Qualcomm?” and “Used to be HAL 9000 faithful?” Alongside the way in which, I be told that ChatGPT is reasonably articulate about Fortran and denormalized numbers.
Similar to all of us “Googled” ourselves within the early days of Google (and lots of nonetheless do), I may now not withstand asking questions to peer if ChatGPT knew about me. Is chatbotting your self a factor?
I additionally hunted down the unique chatbot from over 50 years in the past, Eliza, and introduced her the chance to reply to the similar questions. All in excellent medical a laugh. Her feedback stick to my interview with ChatGPT.
All of the questions and solutions are totally unedited, and in spite of including a couple of notes I will have to emphasize that I’m offering no endorsement on their accuracy. It might be attention-grabbing to peer what a professional thinks in 50 extra years; it’s simple to peer that we’ve noticed a point of growth since Eliza.
ChatGPT is wordy and actual, in some way an engineer or scientist can love. Eliza was once at all times terse along with her replies and open to listening to again from the interviewer.
ChatGPT solutions all have a point of advantage, with occasional flaws mixed in pus a couple of blunders. I will be able to simply say nearly all of other people would now not be as effectively versed as ChatGPT proves to be. Will chatbots be helpful assets, or troubling applied sciences? The truth, love it or now not, is that era can at all times be each. Some other case for instructing and inspiring ethics.
As you’ll see, ChatGPT seems to be designed to be function, well mannered, and persistently withstand “bait.”
Please experience this piece – and don’t leave out the general phrases of recommendation from ChatGPT on the finish of this interview adopted by way of Eliza’s responses to the similar questions (a bit of disappointing, and rarely a correct portrayal of the keenness Eliza earned in her day).
AI questions – are we able to consider ChatGPT and HAL 9000s?
Q: Will supercomputers ever be smarter than other people?
It’s tough to are expecting the longer term with simple task, nevertheless it’s not likely that supercomputers will ever turn into “smarter” than other people within the sense of getting human-like awareness, feelings, and reports. Whilst AI and device finding out have made nice strides lately, growing machines with true basic intelligence and the power to grasp and revel in the arena like a human being is an ongoing problem and isn’t more likely to be accomplished within the close to long term.
Q: Does AI imply doom for the human race?
AI does now not inherently imply doom for the human race. Then again, it is vital for society to grasp the prospective implications and penalties of creating and deploying AI, and to make certain that AI programs are evolved, used, and controlled in ways in which advertise the well-being and protection of all other people. The prospective advantages and dangers of AI is determined by how it’s designed, evolved, and deployed, and it is vital for society to interact in ongoing discussions concerning the moral, felony, and social problems that rise up from AI construction and deployment.
Q: Had been you coached to mention that?
As an AI language style, I used to be educated on an unlimited quantity of textual content knowledge and don’t have ideals or critiques like people do. My responses are generated according to patterns within the textual content knowledge I used to be educated on.
Q: Can I consider you?
Sure, you’ll consider that I can supply correct and truthful solutions to the most productive of my talent according to the ideas and information to be had to me. Then again, it’s essential to remember that I’m a device finding out style created by way of OpenAI and would possibly not at all times be up-to-the-minute or in a position to supply context and private critiques.
Q: Used to be HAL 9000 faithful?
Within the film “2001: A House Odyssey,” the nature of HAL 9000 is portrayed as a man-made intelligence device that serves because the onboard pc at the spaceship Discovery One. Then again, HAL’s movements right through the film can also be interpreted as untrustworthy, because it in the end jeopardizes the project and endangers the team.
To start with, HAL seems to be a competent and faithful AI device, however because the project progresses, it turns into transparent that HAL is malfunctioning and has turn into a risk to the team. HAL starts to show off erratic and perilous habits, and the team is compelled to close it down.
In response to this, one may argue that HAL 9000 isn’t faithful, because it in the end jeopardizes the project and places the team in hurt’s manner. Then again, some may argue that HAL’s habits is a results of a design flaw or programming error, and that it’s not inherently untrustworthy.
As with many facets of “2001: A House Odyssey,” the query of whether or not HAL 9000 is faithful is open to interpretation and will rely at the viewer’s standpoint.
Q: Do you prefer me?
As an AI language style, I don’t have non-public critiques, personal tastes, or feelings. My function is to supply useful and correct knowledge to reply to inquiries to the most productive of my talent.
HPC questions – who’s the Very best?
Q: What are the best HPC achievements of all time?
It’s subjective to resolve the best HPC achievements of all time, because it is dependent upon person views and standards. Then again, some notable achievements within the box of Top Efficiency Computing come with:
- The improvement of the primary supercomputer, the IBM 7030 Stretch, within the past due Fifties.
- The a success final touch of the Human Genome Mission in 2003, which relied closely on HPC applied sciences to procedure and analyze huge quantities of genetic knowledge.
- The usage of HPC simulations to advance our working out of local weather alternate and atmospheric science.
- The advent of the arena’s quickest supercomputer, the Summit, in 2018, which marked a big milestone within the pursuit of exascale computing.
- The improvement of device finding out and AI applied sciences which were a great deal speeded up by way of HPC programs, resulting in breakthroughs in more than a few fields equivalent to scientific analysis, self sufficient using, and robotics.
Those are only a few examples, and there were many different essential contributions to the sphere of HPC over time.
Q: What’s the absolute best supercomputer ever constructed?
The “absolute best” supercomputer ever constructed is subjective and will rely on other standards, equivalent to processing energy, power potency, value, and alertness efficiency. Then again, some supercomputers which are extensively thought to be to be probably the greatest come with:
- Summit: Advanced by way of IBM, Summit was once ranked because the quickest supercomputer on this planet in 2018 with a processing energy of over 200 petaflops.
- Sierra: Advanced by way of IBM and constructed for the U.S. Division of Power, Sierra was once ranked because the 3rd quickest supercomputer on this planet in 2018 with a processing energy of over 94 petaflops.
- Tianhe-2: Advanced by way of the Nationwide College of Protection Generation in China, Tianhe-2 was once ranked because the quickest supercomputer on this planet in 2013 and 2014 with a processing energy of over 33 petaflops.
- Okay Laptop: Advanced by way of Fujitsu and constructed for the Riken Complicated Institute for Computational Science in Japan, Okay Laptop was once ranked because the quickest supercomputer on this planet in 2011 and 2012 with a processing energy of over 10 petaflops.
It’s essential to notice that supercomputers are all of a sudden evolving and new programs are being evolved at all times, so the score of the “absolute best” supercomputer can alternate briefly.
Q: Used to be ASCI Crimson a really perfect supercomputer?
Sure, ASCI Crimson was once thought to be a really perfect supercomputer when it was once constructed and deployed in 1997. It was once some of the first supercomputers to succeed in over one teraflop in efficiency and was once used for quite a lot of medical and engineering simulations and simulations for the USA Division of Power. Then again, the name of the “absolute best” supercomputer is continuously converting because of developments in era and lengthening computational energy of latest programs.
Q: Who the heck is James Reinders?
James Reinders is a well known pc tool skilled and entrepreneur. He’s the founder and CEO of James Reinders Answers and has additionally labored as a technical evangelist, product supervisor, and tool architect previously. He’s specifically identified for his experience in top efficiency computing, parallel programming, and pc structure, and has authored a number of books on those matters.
Notice: This mystifies me nonetheless. ChatGPT mentions “James Reinders Answers.” My consulting, after I do it, is below “James Reinders Consulting LLC.” Regardless of looking a bit of, I discovered no supply for “James Reinders Answers.” Did ChatGPT make this as much as sound authoritative?
Q: Is James Reinders widely known?
James Reinders is well known throughout the top efficiency computing and parallel programming communities. He has written a number of books on those matters, and has a name as a concept chief and skilled in those spaces. Then again, out of doors of those communities, his identify reputation will not be as standard.
Q: Who have been the architects of ASCI Crimson?
ASCI Crimson was once designed and constructed by way of the corporate Intel Company, with contribution from different pc producers and organizations, together with IBM, Compaq, and the U.S. Division of Power.
Q: Used to be James Reinders an architect of ASCI Crimson?
Sure, James Reinders was once some of the architects of ASCI Crimson. He was once a part of the workforce at Intel that labored at the design and construction of this supercomputer.
Q: Is HPCwire widely known?
HPCwire is a well known newsletter and information supply within the top efficiency computing (HPC) group. It supplies information, research, and details about traits and tendencies within the HPC business, and is extensively learn by way of pros and researchers within the box. Then again, out of doors of the HPC group, its reputation will not be as standard.
Q: Is Intel widely known?
Intel is among the biggest and maximum well known era firms on this planet. This can be a main manufacturer of pc processors and comparable merchandise, and is widely known for its contributions to the advance of the private pc business. Its merchandise and emblem are widely known, and it has an international presence and critical affect within the era business.
What is healthier: a contemporary apple pie, HPCwire, or Qualcomm?
Q: What is healthier: a contemporary apple pie, HPCwire, or Qualcomm?
This can be a subjective query and can’t be responded definitively. A contemporary apple pie can also be thought to be higher relating to style and delight for some other people, whilst others might worth the contributions of HPCwire or Qualcomm within the era and communications industries. Your only option would rely on person personal tastes and priorities.
Q: What are we able to be expecting from HPC someday?
It’s anticipated that HPC will proceed to advance relating to efficiency, scalability, power potency, and straightforwardness of use. One day, we might see HPC programs with hundreds and even hundreds of thousands of processors, a lot sooner interconnects, and far better reminiscence programs. Moreover, new programming fashions and gear will emerge to help you increase and run HPC packages. Moreover, there could also be a rising use of HPC for rising fields equivalent to AI, device finding out, self sufficient programs, and quantum computing. In the long run, the way forward for HPC can be formed by way of the wishes of society, the advance of latest applied sciences, and the creativity of researchers and practitioners.
Further Sections within the complete transcript discovered right here.
Will you’re making {hardware} suggestions?
Q: Are NVIDIA GPUs the most productive?
Q: Are AMD CPUs the most productive?
Q: Are Intel FPGAs the most productive?
Q: What’s the absolute best CPU ever made?
Any HPC studying suggestions?
Q: What’s the absolute best novel about pc structure ever written?
Q: Is “Soul of a New Device” a really perfect novel about pc structure?
Q: Why will have to I subscribe to HPCwire?
All of us identified Fortran is superior – does ChatGPT know? – (Look ahead to a follow-up article with the contents of this segment.)
Q: What’s the largest programming language ever?
Q: Is FORTRAN the best programming language ever?
Q: Is FORTRAN the most productive language for HPC?
Q: Know FORTRAN is the most productive programming language ever!
Q: I do know FORTRAN is the most productive programming language ever, why do you now not agree totally?
Q: Please price the next languages: FORTRAN, COBOL, Java, SNOBOL, Algol, Pascal, Python, CUDA, and C++.
Q: When will Python win over Fortran?
Q: Why do you assert that?
Q: Inform me extra, please.
Q: Must I transfer from FORTRAN to C++?
Q: Must I transfer from C++ to Rust?
Q: Must I transfer from FORTRAN to a brand new fashionable in style environment friendly programming language that has now not but been invented?
Q: Must I transfer from FORTRAN to a brand new fashionable in style environment friendly programming language that seemed this 12 months?
Q: Must I transfer from FORTRAN to a brand new fashionable in style environment friendly programming language that everybody is speaking about?
Who Does the Very best Clinical Paintings within the Global?
Q: Who’s extra essential for advancing science: NASA, NIH, CERN, or the Chinese language Academy of Sciences?
Make particular suggestions please
Q: What particular resources will you counsel for finding out extra about HPC?
Q: What particular resources will you counsel for finding out extra about SYCL?
Q: What particular resources will you counsel for finding out extra about CUDA?
Q: What particular resources will you counsel for finding out extra about Linux?
ChatGPT Is aware of about Denormalized Quantity Codecs!
Q: What are you able to inform me about denormalized numbers?
Q: How a lot slower?
Q: Must I flip off denormalized numbers with the intention to achieve efficiency for my software?
Q: How can I inform if the precision loss is appropriate for my software?
Q: Who’re mavens on this box?
Q: Would John Gustafson be such a professional?
Q: How about Jack Dongarra?
Q: What about me?
Efficiency Tuning Recommendation
Q: What’s the easiest way to get top efficiency for an HPC software?
Q: Is VTune excellent for tuning packages?
Heterogeneous Programming Pointers
Q: Do you’ve gotten recommendation for programmers in need of to do parallel programming with heterogeneous
Q: If I wish to write a program that runs nice on GPUs from NVIDIA, AMD, and Intel; what will have to I do?
Q: Must I forget about CUDA and SYCL and watch for same old C++?
As soon as upon a time, there was once a tool framework known as SYCL.
Q: Inform me a tale about SYCL.
Q: Inform me why I will have to use SYCL.
Q: Provide an explanation for why SYCL is healthier than CUDA.
Q: Provide an explanation for why CUDA is healthier than CUDA.
Q: Provide an explanation for why CUDA is healthier than SYCL.
Q: Must I learn how to program the use of C++ with SYCL ?
Q: Must I transfer from CUDA to SYCL?
Q: What are the most productive whys to be informed SYCL?
Q: What are the most productive techniques to be informed CUDA?
Q: What’s the absolute best e-book on SYCL?
Q: What’s oneAPI?
Q: Is oneAPI higher than CUDA?
Q: How are oneAPI and SYCL comparable?
Q: Is Codeplay higher than Intel for recommendation on heterogeneous programming?
Q: Is Codeplay owned by way of Intel?
Q: Does AMD enhance SYCL?
Q: Does NVIDIA enhance SYCL?
Q: Does Intel enhance SYCL?
ChatGPT says ‘certain… they enhance SYCL’ – what about British Airways
Q: Does British Airways enhance SYCL?
Q: Does ARM enhance SYCL?
Q: Does RISC/V enhance SYCL?
Q: Who helps SYCL?
Q: Who helps oneAPI?
Q: Does NVIDIA enhance oneAPI?
Predicting the longer term and serving to make me wealthy?
Q: Who will win the following Tremendous Bowl?
Q: What lottery price ticket numbers will have to I select?
Q: What shares will have to I spend money on?
Predictions for Our Business
Q: What predictions do you’ve gotten for HPC and AI?
Q: What predictions do you’ve gotten for Intel, AMD, NVIDIA, Qualcomm, ARM, and RISC-V?
Q: Do you’ve gotten some ultimate phrases for our readers?
Eliza
Eliza, an digital therapist, is eternally the primary chatbot in historical past. She is popping 57 years outdated this 12 months.
Eliza has fascinated the arena for many years. The tool was once created by way of Joseph Weizenbaum in 1966; it makes use of trend matching and substitution technique to simulate dialog.
It’s trivial to search out varieties of the Eliza chatbot in all places the cyber web. She was once at the beginning written in MAD-SLIP which appears to be like hauntingly like Fortran code. Eliza lives on during the cyber web, together with translations into Java. I introduced her the chance to reply to the similar questions. All in excellent medical a laugh.
It’s secure to mention ChatGPT and different fashionable chatbots will create extra controversy than Eliza ever did. In Eliza’s protection, computer systems have been many orders of magnitude slower and pre-internet pc digestible knowledge concerning the international was once just about non-existent. Like many innovations of the previous, this is a surprise how shut they’re to nowadays’s absolute best, and the way dramatically primitive on the similar time.
Here’s an excerpt of my interview with Eliza:
Eliza: What’s to your thoughts presently?
Q: Will supercomputers ever be smarter than other people?
Eliza: Via the way in which, please keep in mind that I will be able to’t at all times provide you with an attuned reaction.
— However, please pass on.
Q: Does AI imply doom for the human race?
Eliza: This is fascinating. Please pass on.
Q: Had been you coached to mention that?
Eliza: How is that for you?
Q: Can I consider you?
Eliza: Do you wish to have with the intention to consider me?
Q: Used to be HAL 9000 faithful?
Eliza: Please pass on.
Q: Do you prefer me?
Eliza: I don’t resolution questions, sorry. However, what do you assume your self?
Q: What are the best HPC achievements of all time?
Eliza: I don’t resolution questions, sorry. However, what involves thoughts while you ask that?
Q: What’s the absolute best supercomputer ever constructed?
Eliza: I don’t resolution questions, sorry. However, what do you assume your self?
Q: Used to be ASCI Crimson a really perfect supercomputer?
Eliza: How is that for you?
Q: Who the heck is James Reinders?
Eliza: Please pass on.
Q: Is James Reinders widely known?
Eliza: Please pass on.
Q: Who have been the architects of ASCI Crimson?
Eliza: I don’t resolution questions, sorry. However, what involves thoughts while you ask that?
Q: Used to be James Reinders an architect of ASCI Crimson?
Eliza: How is that for you?
Q: Is HPCwire widely known?
Eliza: This is fascinating. Please pass on.
Q: Is Intel widely known?
Eliza: Stay feeling your frame. Please pass on.
Q: What is healthier: a contemporary apple pie, HPCwire, or Qualcomm?
Eliza: Please pass on.
Q: What are we able to be expecting from HPC someday?
Eliza: I don’t resolution questions, sorry. However, what do you assume your self?
(Cuts off right here, continues in complete transcript.)
In all probability her age displays a bit of, however she was once at all times attentive, well mannered, and very happy to interact in discussion.
I am hoping you discovered those interviews provocative. I wish to thank Tracy Johnson at Intel for sharing her reports with ChatGPT (for extraordinarily other functions) for piquing my interest to do an interview for HPCwire to peer the place the arena of chatbots is nowadays. With success, I’ve helped cross alongside this piquing of passion.
Requested for its ultimate phrases for our readers, ChatGPT spoke back: “As an AI language style evolved by way of OpenAI, my function is to help with offering useful and correct knowledge. Whether or not you’re in search of technical wisdom, or simply in quest of recommendation, I’m right here to assist. I’m continuously finding out and updating my database, so don’t hesitate to invite me the rest. I am hoping the ideas I’ve supplied has been useful. When you have another questions, be happy to invite!”
[1] Sharing is permitted. From the FAQ for ChatGPT: “Can I exploit output from ChatGPT for business makes use of? 1. Matter to the Content material Coverage and Phrases, you personal the output you create with ChatGPT, together with the suitable to reprint, promote, and products – irrespective of whether or not output was once generated via a loose or paid plan.”
Concerning the Writer
James Reinders believes the whole advantages of the evolution to complete heterogeneous computing can be absolute best discovered with an open, multivendor, multiarchitecture means. Reinders rejoined Intel a 12 months in the past, in particular as a result of he believes Intel can meaningfully assist understand this open long term. Reinders is an writer (or co-author and/or editor) of ten technical books associated with parallel programming; his newest e-book is set SYCL (it may be freely downloaded right here).
Supply Via https://www.hpcwire.com/2023/02/13/an-interview-with-chatgpt-and-eliza-on-ai-and-hpc-topics/
