Microsoft Going All-in On AI
The future is here, whether we like it or not.
Bloomberg (“Microsoft to Bring OpenAI’s Chatbot Technology to the Office“):
Microsoft Corp.’s effort to overhaul its entire lineup with OpenAI technology has spread to one of the company’s oldest and best-known products: its Office apps.
The software, including Excel, PowerPoint, Outlook and Word, will begin using OpenAI’s new GPT-4 artificial intelligence platform, Microsoft said on Thursday. AI-powered assistants called Copilots will be able to generate whole documents, emails and slide decks from knowledge the software has gained scanning corporate files and listening to conference calls. The technology will debut in the coming months, and Microsoft is already testing it with 20 companies, including eight in the Fortune 500 that it declined to name.
“This is the big next step for us — to put it in the tools everybody uses every day for their work,” Chief Executive Officer Satya Nadella said in an interview. The new technology will help people create “great content, great documents, great PowerPoints, art,” he said, as well as do sophisticated analysis using natural language queries.
The move is part of a stampede of companies adding AI chatbot features to their technology. OpenAI, backed by Microsoft, has fueled much of the frenzy with its ChatGPT tool, which went viral in recent months and demonstrated the power — and potential pitfalls — of chatbot technology. The startup just unveiled GPT-4, the latest iteration of the underlying software, earlier this week.
Microsoft has already been using the system in its Bing search preview for several weeks. After several reports that the included chatbot was generating freewheeling conversations that some found strange or belligerent, the company began restricting its responses.
Microsoft, which is investing more than $10 billion in OpenAI, also has released Copilot software for sales and customer applications, as well as a product from its GitHub unit for writing programming code. And its biggest rival in office software, Alphabet Inc.’s Google, announced its own plans this week to use AI tools for things like creating presentations, taking notes during meetings and drafting emails.
Microsoft-owned LinkedIn, meanwhile, said on Wednesday that it will use OpenAI’s language technology to help users create a more compelling profile on the professional networking site.
The Verge (“Microsoft Business Chat is like the Bing AI bot but as a personal assistant“):
One of the new Copilot AI features coming to Microsoft 365 apps and services is dubbed Business Chat. It’s a chatbot experience that’s able to summarize information pulled from meeting transcripts, recent contacts with customers, entries in your calendar, and more that you can plug into emails for the team or as slides in a presentation.
According to Microsoft, by using grounding to focus the AI on your business’ trove of data, it can create relevant, accurate responses to natural language prompts, like “Did anything happen yesterday with [customer X]?” The bot is accessible from Microsoft365.com, Bing when signed in with a work account, or via Microsoft Teams.
The output will look familiar if you’ve seen the Bing AI chatbot at work, plugging in footnotes to show where it obtained particular data, and in the demo, there was a focus on how users can update or correct entries as necessary.
There’s an acknowledgment in the pitch that it may not create a perfect finished product every time — we’ve seen the potential flaws of chatbot AI already. But this isn’t just a time saver; Microsoft suggests you can ask it to brainstorm on flaws in a strategy and even ways to deal with them.
Here’s a list of proposed prompts:
- Summarize the chats, emails, and documents about the [customer] escalation that happened last night.
- What is the next milestone on [project]? Were there any risks identified? Help me brainstorm a list of some potential mitigations.
- Write a new planning overview in the style of [filename A] that contains the planning timeline from [filename B] and incorporates the project list in the email from [person].
The new Copilot AI features are in limited testing right now with 20 of Microsoft’s customers, and details about pricing haven’t been announced. The company says, “We will be expanding these previews to customers more broadly in the coming months and will share more on new controls for IT admins so that they can plan with confidence to enable Copilot across their organizations.”
CNN Business (“Microsoft is bringing ChatGPT technology to Word, Excel and Outlook“):
Microsoft on Thursday outlined its plans to bring artificial intelligence to its most recognizable productivity tools, including Outlook, PowerPoint, Excel and Word, with the promise of changing how millions do their work every day.
At an event on Thursday, the company announced that Microsoft 365 users will soon be able to use what the company is calling an AI “Co-pilot,” which will help edit, summarize, create and compare documents. But don’t call it Clippy. The new features, which are built on the same technology that underpins ChatGPT, are far more powerful (and less anthropomorphized) than its wide-eyed, paperclip-shaped predecessor.
With the new features, users will be able to transcribe meeting notes during a Skype call, summarize long email threads to quickly draft suggested replies, request to create a specific chart in Excel, and turn a Word document into a PowerPoint presentation in seconds.
Microsoft is also introducing a concept called Business Chat, an agent that essentially rides along with the user as they work and tries to understand and make sense of their Microsoft 365 data. The agent will know what’s in a user’s email and on their calendar for the day as well as the documents they’ve been working on, the presentations they’ve been making, the people they’re meeting with, and the chats happening on their Teams platform, according to the company. Users can then ask Business Chat to do tasks such as write a status report by summarizing all of the documents across platforms on a certain project, and then draft an email that could be sent to their team with an update.
Microsoft’s announcement comes a month after it brought similar AI-powered features to Bing and amid a renewed arms race in the tech industry to develop and deploy AI tools that can change how people work, shop and create. Earlier this week, rival Google announced it is also bringing AI to its productivity tools, including Gmail, Sheets and Docs.
The news also comes two days after OpenAI, the company behind Microsoft’s artificial intelligence technology and the creator of ChatGPT, unveiled its next-generation model, GPT-4. The update has stunned many users in early tests and a company demo with its ability to draft lawsuits, pass standardized exams and build a working website from a hand-drawn sketch.
OpenAI said it added more “guardrails” to keep conversations on track and has worked to make the tool less biased. But the update, and the moves by larger tech companies to integrate this technology, could add to challenging questions around how AI tools can upend professions, enable students to cheat, and shift our relationship with technology. Microsoft’s new Bing browser has already been using GPT-4, for better or worse.
A Microsoft spokesperson said 365 users accessing the new AI tools should be reminded the technology is a work in progress and information will need to be double checked. Although OpenAI has made vast improvements to its latest model, GPT-4 has similar limitations to previous versions. The company said it can still make “simple reasoning errors” or be “overly gullible in accepting obvious false statements from a user,” and does not fact check.
Still, Microsoft believes the changes will improve the experience of people at work in a significant way by allowing them to do tasks easier and less tedious, freeing them up to be more analytical and creative.
The technology is in its early days still and I don’t make any claim to fully understanding any of it.
I’ve played around with ChatGPT a bit, wondering if it could replace bloggers. Its integration into Microsoft and Google’s applications makes it more plausible; I’d be surprised if WordPress doesn’t follow suit. While I don’t think it will replace the analysis that the best bloggers bring to the table, it could certainly speed up the tedious work of linking to and aggregating sources to make getting to that part of more efficient.
Those of us in the academy have been wrestling with the implications for student evaluation. If artificial intelligence—or machine learning, or whatever you want to call it—is already at the point where it can mimick an average student response to an example question or essay prompt, how do we handle that? Do we try to “ChatGPT-proof” our questions? Or do we simply treat that technology as another tool students can use—akin to spelling and grammar checkers or bibliography-generating tools—and bake it into the assignment structure somehow? If even Microsoft Word has that technology built in, the latter would seem almost required.
Dear technology: Stop being “helpful”!!
One of the first things I do when I get a new computer or phone (or just apps) is turn off all the “Let me help you” crap. Even auto-complete* gets fracking annoying. I want my machines to do what I tell them to. If I make a mistake, that’s on me.
* Dear auto-complete: It’s never “duck”. Never.
@Mu Yixiao: I’ve (mostly) learned to reread my comments here to see if anything was “helpfully” changed between typing and posting. And it can be a real struggle to deliberately misuse a word. I’m with you, I want things to be deterministic. If I input X I want to see output Y every time. I finally figured out how to turn off the “smart” cruise control on my newish Honda so it just, you know, controls cruise speed without constantly annoying me by trying to be “smart”.
And Microsoft is listening to conference calls? Also too, IF MS is automating office work we’d better be looking at UBI.
Sometimes it is.
I’m sure looking up a recipe for Peking f*ck would result in a ducking mess.
Siri ended up being nothing but a kitchen timer. The Apple watch is a pedometer. My smart home is a pain in the ass my daughter has to fix whenever she visits. My security camera likes to tell me I have a package when I don’t. I half suspect this will be similarly disappointing. Or it may be the beginning of the end of humanity.
A Peking fuck is what the Beijing Craig’s List personals section is for. 😀
If the AI writes the blog piece would the commenters also be AI? People would just sit around and read blogs and get angry without having to do anything?
Humans are constantly rushing to embrace new technology before they’ve considered all the possible consequences of the change. I’m sure it goes all the way back to Ogg showing his latest genius creation to his cavemates who aren’t impressed. Hopefully we haven’t dorked ourselves with this one.
AI will be a god send to those unlucky corporate schmucks that get dozens, if not hundreds of emails per day, but they are too low on the totem pole to simply ignore them.
It’s always hard to tell the effect a new technology will have. I am reminded that the concept of the “tablet computer” originated in the 80’s at Xerox PARC and Microsoft had three attempts at it, with entirely new operating systems each time and billions of dollars in investment, all failures. And then Apple introduced the iPad and now they are so ubiquitous we take them for granted. But there is one thing about this new “AI” technology I’m pretty certain about: the advice I’ve given my kids since they were little is about to get blown up. I’ve long told them that, unlike their parents, they are living in the biggest small town that ever existed. Anyone, anywhere can know every public thing they ever did. In a small town in the 1950’s, if three 13 year olds sampled some whiskey out of an unlocked liquor cabinet and ended up running naked into the corner store and crashing into a display of canned peas, well, it would be the thing that everyone in town knew about those kids, even decades later. But if they moved to a new town a state or two away, no one there would ever know. But the Web brought that 13 year old’s story to everyone, everywhere (all at once).
Fundamentally, what I told them was that anyone, anywhere could know everything you’ve ever done in public and a lot of what you’ve done in private. While that is still true, what is going to change is that anyone, anywhere can know a whole lot of things you HAVEN’T done, so many that nothing online will be trusted. Even today, if some random high schooler wants to see that new girl in geometry class naked he doesn’t have to go searching to see if a sext has made it into the wild, he can simply go to a porn site, pick a performer who resembles her, supply a bunch of pictures of her face, and voila, for all intents and purposes a porn video of that girl. A year from now and he wont even have to do that much work, he can just describe what he wants to see and who with and the porn AI will supply it. And of course he will share it with his friends and, well, you get the picture. And this applies to anyone. Do you want a grainy phone video of Joe Biden telling someone what he really thinks about Kamala Harris or that he is secretly plotting to take everyone’s guns? Here’s twenty of them.
We will be quickly getting back to the days when secrets can be kept, because a real video of Matt Gaetz shagging a 16 year old will be buried amongst a thousand AI generated ones.
The whole tech world utopia has been reduced to something touted in a 3-day seminar at a Four Seasons in Tuscon. I’m not saying the dreams/dystopias of the 80s and 90s were in any way realistic. But going from Skynet and Neuromancer to an AI that can draft a lawsuit…This is like a joke in a Philip K Dick short story: they made the ambulance chasers robots…
I imagine that the flaws in handing the keys to AIs are going to be funny. Like driverless cars which need to be told that people can walk outside a crosswalk, there’s going to be a funny mistake which causes some financial damage. But overall, this stuff seems like it will be absorbed into the background noise of life.
There’s an urgent need for a super-powerful AI that can accomplish one particular task: turning a PDF document into a legible, ordered Word document that keeps the original format and does not add line or page breaks.
It should actually be easy. No one composes documents in PDF, they all use Word or some other word processing program. Then they save the .docx file as a .pdf. But you can’t change the .pdf back into .docx.
Oh, you can convert it to .docx, but the result is more an interpretation than an accurate, functional copy of the file. You’d think the .pdf could keep an image of the original Word and reconvert to .docx well.
In the past three or four years, we’re getting more files from government agencies as PDF. Often there are large amounts of data that need to be transcribed in the proposal. these tend to have lots of tables and other formatting.
When I get them in Word, it’s a short, straightforward process of copy-paste, with little to no reformatting required later. Copying from PDF direct nets you a mess that looks nothing like the original. Converting to Word nets you a similar looking file that won’t copy well and needs a lot of reformatting and attestation to detail. TL;DR if I get a word document 200 pages long, I can incorporate it as needed to the proposal in 30-45 minutes; the same document in PDF takes severa hours, sometimes even two days.
Prediction… Grades will become mostly dependent on work done in a classroom or large test room, with no electronics allowed, in all lower level classes, not just math and science. ChatGPT and its ilk are only the newest thing; Wolfram’s Mathematica, for example, has been better than humans at derivatives and integrals for many years. In the humanities, students not fluent in cursive will be at a significant disadvantage.
Perhaps of interest, just as ChatGPT uses methods that humans can’t, Mathematica’s technology for doing derivatives and integrals is also something that humans can’t mimic.
Go old school – tests and essays done in the classroom and written on computers not connected to the internet.
@Kathy: PDF, like PostScript before it, is a page description language. That is, for example, “Put this character at that position on the page using a particular font, size, color, and rotation.” Going backwards from that to any format based on things like paragraphs, lists, headings, captions, etc is a hard problem. Recognizing a hyphen at the end of a line of text is one thing. Determining accurately whether it’s there to indicate a non-hyphenated word split across lines versus a break in a hyphenated compound word is non-trivial. If it were easy, or even straightforward, OCR-based book scanning would be much more accurate than it is.
Next step, wearables/active contact lenses and then implants.
Classroom Faraday cages?
Next step after that: implanted AI supercomputer with cached significant useful fraction of the non-video internet.
“That’s my secret , Cap. I’m always angry.”
@steve: AI would also do the reading and getting angry as well.
I haven’t turned off autocorrect on my phone, I have instead learned how to mostly leverage it so that it is on balance a plus. I have to watch it every second though. What I would like, and what I will probably never get, is one that learns what I type, rather than using the patterns it sees in general.
Interestingly enough, when I do a Google search on their website, it has an autocomplete that does seem to remember what I’m interested in. But not the text app on my phone, alas.
One of my biggest concerns overall is that the ai will be tuned to serve the interests of the people who paid for it. So if I have Microsoft’s Chatbot write a 5-page report for me, it might well be that it inserts an ad for Microsoft, or maybe for someone who has paid Microsoft. Or maybe it just reinforces the value of (spreadsheets, word processors, or chatbots!). I mean, Facebook’s Algorithm serves them, not me, and likewise YouTube.
Here’s another prediction. A number of years ago, someone (the author of “Bowling Alone”?) was commenting on the effect that TV had on socialization. He observed that your “TV friends” were cooler, funnier and better looking than your real life ones, never got mad at you, didn’t care if you ignored them for months and then rejoined them as if there had been no interruption. He believed that TV friends fulfilled a non-negligible fraction of people’s need for socialization. So – how much larger will the fraction be when those “TV friends” are ChatBot friends, who are cooler and funnier than your real ones, better looking and massively interactive? They will be willing to talk about whatever you want, as deeply as you want, and never get bored or tired.
So here’s my prediction(s): Within five years you will come across articles about people who have developed an odd and perhaps unhealthy relationship with a ChatBot, in much the same way as those quirky Japanese men “marry” an anime character. I can see lawsuits if a company changes or discontinues a specific interactive character. And within 10 years, 15 at the most, there will be a vocal group of people that believe they are alive. How’s that for going out on a limb, prediction-wise?
Postscript: When I wrote my predictions I had in mind a story of a Japanese man that had “married” a body pillow with an anime character printed on it. Just now I googled to try to find a reference, and I found this. It turns out that in Japan there is at least one guy who “married” an anime ChatBot in 2017, but the company that hosted her has discontinued the character. No lawsuit, but he insists that he will remain faithful to her until he dies, even though he cannot talk to her anymore.
Trivia: the x in docx stands for xml. You can change the file extension to .zip, extract all, and poke around in the underlying storage format.
At which point the academic world will have had to figure out that they either teach something that the software can’t do, or they’re superfluous. The last time I taught Calc I at the local community college, one of my in-class tests was all word problems. With instructions to set up the problem but not solve it.
I know someone who takes great joy in getting ChatGPT to do odd things. I’ll have to ask him to run some calculus word problems through it.
I’d a vague notion of the issues involved.
It reminds me of an anecdote about early electronic translators that came out in the 80s. someone tried translating “Out of sight, out of mind” to another language. Then took the result and translated it back to English. What came out was “Invisible, insanity.”
The thing is that PDF was meant as a universal format, with free Acrobat PDF reader available. Since now Word is ubiquitous, and even other office suites can display it, PDF is no longer necessary for documents meant to be edited or copied.
Or, given how much cheap memory is available these days, not to mention broadband transmission speeds, Word files converted to PDF could contain the original .docx code for converting back to Word.
That sounds well above my paygrade.
Of course in a world where a machine does the computation far more efficiently than a human ever can and AI can have the capability to choose the data set and the proper equations to analyze that data based on a humans request, what value does knowing how to perform those tasks have? What becomes most important, and was really what has always been important knowing how to interpret the output.
@Michael Cain: An interesting approach to Calculus, for sure. I just watched a great video demonstrating how Archimedes did calculus without knowing any algebra. He just did it geometrically. Kind of amazing.
I think that the value people will find in this, in academia, is in that people see value in understanding more about the world they live in, and calculus is a magnificent tool for that. Calculus instruction has be very focused on practical matters – we need to get these engineers up to speed so that they can understand, frame (as in your exam) and solve problems.
But maybe only the first two will be important as we go on. When I took differential equations, the focus was on technique (excluding the LaGrange transform and that whole routine!). Which is important, but allowed me to miss the big picture – which was that most of how we model much of anything in the world involves the solution of a set of pdes, mostly by numerical methods. Math classes don’t tend to discuss that, as math is imagined to be interesting in its own right. (I’ve been guilty of that).
Here’s a good use for that. My father has dementia and is in a home in San Diego. Neither of his kids, nor any of his grandkids, live nearby. Frankly the man is not a great conversationalist these days, what with repeating the same (false) stories over and over again. But an AI could patiently reply with the obligatory, “huhs” and, “you don’t says.” We’ve got a hell of a lot of Boomers heading down that path and not enough willing workers to care for them.
But I suspect AIs won’t take over the creative world. I’ve found that special effects, computer graphics, all things green screen, no longer work for me. I increasingly see them as ones and zeros, a magician’s trick, false. I’ve come to actively resent them. Show me a computer representation of a train hitting a car and I just don’t care. Do it as a practical effect? Somehow that just feels different. IOW, an AI’s game may be limited by the fact that humans simply won’t relate emotionally precisely because they know it’s an AI.
There have been trials of robotic helpmates for the elderly and one of the results that was a bit of a surprise, is that the elder began to address the robot as if it were a person, similar to how people interact with pets.
You know, I understand your point. I agree with you in a very general sort of way. However, I also note that that is quite a thing for someone who writes fiction (also known as “telling lies”) for a living to say.
(Did the twinkle in my eye come through?)
Also, I love magicians tricks.
@Sleeping Dog: The Japanese have funded this heavily. Being a society where racial purity is of utmost importance, they are very reluctant to let low level foreigners in, even as temporary workers. And because their population is aging quickly and their birthrate so low, they have more and more mentally and physically infirm elders and fewer and fewer people left to care for them. They hope they can supplement significantly with robots in all kinds of roles, from conversational companion to nursing aide capable of helping someone up off the floor and calling for help if necessary.
@Jay L Gischer:
The “standard” math sequence from high school algebra through a first semester of differential equations and linear algebra is a tough call. On one hand it is, as you say, part of the “language” in many technical fields. On a different hand, almost no one solves things by hand any more. (Back when I was a supervisor, if someone in my group were solving integrals by hand rather than running them through Mathematica, we would have had a discussion about best uses of their time.) On yet another hand, for people who are going farther in math, it’s one of the first points where they get to think about functions as first-class objects: the solution to this problem is not a number, it’s a function.
I’m usually careful to say that Archimedes — and lots of other ancient people — did things “like” calculus. You can only push those approaches so far. In the period from about 1820-1850, mathematicians tossed all of the old approaches and rebuilt calculus from set theory up, then went on.
1) Word is not ubiquitous. I don’t have a copy because (a) my desktop machine runs Linux, so I would have to keep a Windows virtual machine available to run it, and (b) I choose not to pay the Microsoft Windows tax. Also relevant, MS does not guarantee that Word for Windows and Word for Mac are consistent. Embedded objects like spreadsheets are particularly problematic.
2) While tools like LibreOffice do an impressive job of parsing a docx file, the ECMA spec for the format is 5,000 pages and many items in it are marked as “implementation specific”. There is no guarantee that any particular document will be imported correctly.
3) The details of Word to PDF are in Word’s layout engine, about which MS has revealed almost nothing. It is a sure thing that LibreOffice’s layout engine for PDF will make some different choices in a complex document.
@stevve: I think the bot part of commentary threads will only get larger, but it may still only be artificial lack of intelligence like it is now.
@Andy: Also, ask questions that call for information and analysis (preferably analysis from 2 or more viewpoints*) rather than just information. But yeah, the only test I ever took on a laptop was the reading test for my MA–6 questions three hours to write–minus the time spent getting the fax of the test and sending the fax of the printout back.
*In other words, tests like one of my professors gave where she provided us with a list of 25 or so possible questions, “so we’d know what to study.”
@Kathy: “translating “Out of sight, out of mind” to another language.”
But the result is involved with the nature of idioms in language. Often, I would ask my French teacher “how do you say ‘X’ in French,” and she would reply “you can’t; French people don’t say that.”
@Michael Reynolds: Given how quickly AIs have learned racism in the past, I think we should protect them from the influences of our older generations.
In the spirit of St. Patrick’s Day, I offer this vision. A world with AI police forces that have recognized society’s biases against the darker hues. The Irish reign supreme with their pale, almost literally white skin. England trembles.
Not exactly, but there are enough bots pushing shit, both as original tweets and replies, that without careful pruning it will just become bots arguing with bots, dominating discourse.
Directors do seem to have a contest going to to see who can do the biggest, loudest, most colorful, and in terms of the plot, pointless CGI. They’re supposedly using multi-physics computer models to make things look realistic. It isn’t working.
I was really impressed with Dunkirk. When the director, Christopher Nolan, needed a navy destroyer, he rented one. He needed a thousand troops on the beach, he rented a beach, borrowed some troops, and made a bunch of cardboard cutouts. He needed Spifires and Heinkels and Messerschmitt’s so he hired some from warbirds owners. Many the same planes used in 1969’s Battle of Britain. When he couldn’t fit cameras in a Spitfire he hung cameras on an old Russian YAK-52 training plane and made it a “Yakfire” by dummying up a Spitfire cockpit interior. Made the whole movie much more credible to my eye. OK, it was really more a movie about the British myth of Dunkirk than the actual event, but a credible myth.
MS Word is ubiquitous enough that it makes more sense to have Word files than PDF files if editing or copying is required.
As to embedded objects, I’d just as soon do away with them. There’s nothing worst than having whole files embedded in a document.
@Just nutha ignint cracker:
Then your teachers weren’t very good. The concepts of idioms are fairly universal. And they most certainly translate easily between Latin-derived languages. The idiom exists in French, your teacher just didn’t know how to match it with the English one.
While teaching in China, I would do an entire class on idioms–and work them in to numerous other classes. I don’t think I ever found an English idiom for which there wasn’t a Chinese idiom that expressed the same basic concept*.
Kill two birds with one stone = two birds, one arrow
A drop in a bucket = one hair, nine oxen
In for a penny, in for a pound = don’t do, or don’t rest (not exactly, but close)
In a bubble = view the sky from a well
Like wildfire = like wild fire
Work hand in glove = villains collude together
* Chinese idioms are fun, because they (usually) blatantly announce that they’re an idiom. They are all either 4 or 8 characters (words)
It should be noted that MS Word has serious compatibility issues with files made in MS Word. Cross-platform issues, earlier version issues, etc. Official docs saved in .doc format from a dozen years ago may display completely differently today.
Some years back, a colleague went back to resume work on a textbook he’d been writing in Word several releases previously. The current version of Word wouldn’t open it; MS didn’t have a translation tool; eventually he sent me a copy and whichever of OpenOffice or LibreOffice I was running at the time had no problems with it.
Being a computer file pack rat, the other day I stumbled across some document I’d written 35+ years ago using the troff and friends markup languages. Groff rendered it flawlessly, including the embedded tables and equations. And since it’s a markup language intended to be human-readable in a plain text file, I could have modified it easily using any text editor.
Just a personal opinion, but groff and tbl will produce, first attempt, tables that look far better than what can be done in Word (or raw HTML tables) without a lot of tedious manual formatting. I cheerfully admit that if I’m writing for the Web, I use groff and tbl to generate the table, then take a snapshot from the screen and insert that image rather than mess with the HTML.
Usually far less than with PDF formats.
We had a spot of trouble in the mid-2000s when we still used the .doc rather than .docx format and program. But there was a utility online to make the conversion.
I get it about older versions. I transfer older files to newer versions as they come along. I’ve some files that began life in WordPerfect in the 80s (really).
But I mostly mean work files now, and those are produced year to year. There’s little chance we’ll suddenly find documents in vintage Win2.0 Word versions.
I’ve been playing with ChatGPT at work.
We’re interested in it’s potential in several fields: database query for general users, coding, digitisation of footnote and equation-rich documents (a PITA, let me tell you), and detecting plagiarism and similar fakery.
It is hopeless with mathematics; it looks OK to me, but the problem according to someone who knows, is that is NOT calculating. It’s basing it’s answers on consensus info from the cached internet. When that’s mistaken, so’s ChatGPT. And it even seems to try to cover up it’s misatkes, sometimes. LOL
It has no “truth valorisation”; at present, it’s purely an association engine.
A very good one, but that’s it.
However, people who dismiss it may be missing something. Strap together a “n’th generation” iteration of ChatGPT with other neural networks designed for formal logic etc etc and we may start to see real “Simulated Sentience” systems.
We know hardware evolution of response to environment can drive to high level sentience, and possibly even sapience.
Because, here we are, chatting. 🙂
Here is the thing…
If we take a look at the general idea that Star Trek laid down, it was generally an optomistic view of technology and humanity: Once technology moves forward, then everyone would have plenty without burden, the need for money would end, and people were free to follow their interests that would provide them a fulfilled life.
Not a snowballs chance on hell that will happen here.
Here, we already see (if you haven’t noticed) that every major tech company is laying off massive numbers of people. And the likelihood of those highly paid tech folk finding a job when most every tech and software company continues to downsize and freezes hiring is problematic.
There is no recession, but the bloodletting continues unabated.
People are shown the street, and good luck with that. Folks will find jobs, but at a smaller fraction of their highly paid salaries. Homes will need to be sold; defaults will occur… but still money will flow to the top. Because that is how companies and capitalism is structured.
It’s not Star Trek, it’s the very worst of Elysium. Coming at you fast.
@MarkedMan:..Even today, if some random high schooler wants to see that new girl in geometry class naked he doesn’t have to go searching to see if a sext has made it into the wild, he can simply go to a porn site, pick a performer who resembles her, supply a bunch of pictures of her face, and voila, for all intents and purposes a porn video of that girl.
How times have changed.
Can’t remember how it came up in conversation but about 30+ years ago my neighbor told me about how one of his daughters wanted to pose in Playboy for a feature titled The Girls of University of Illinois in her birthday suit. She told him about it before she did it and dad told her to go for it. Can’t remember how I found out which issue it was but I confess that I surveyed all the used bookstores in town and never could find it. It was a chore looking through dozens of old Playboys but to no avail. Somehow I thought that while dad might given his approval maybe he bought up all those old copies.
Several years later when I was visiting San Francisco I found the issue and sure enough there she was. Naked as a jaybird and smiling at the camera.
(I know U of I lost in their first round game yesterday to Arkansas but what the hell.)
I am old, and remember MS buying up every company with interesting video compression technology in the mid-1990s, setting the field back a decade or more. Being convicted as a monopoly on the basis of Bill’s “Kill Apple” emails. Of making lots of noises but still not conforming to an open standard for .docx formats. I trust them as far as I can throw their headquarters’ building, and will never agree to adopting any of their proprietary formats.
I’m not saying you’re wrong, but when a client with a multimillion dollar contract specifies certain key documents must be uploaded in Word and Excel formats, we’re really not about to use LibreOffice nor any other open source alternative.
And just maybe this is how Gates bought them out