Anyone else witnessing a panic inside NLP orgs of big tech companies?

Wow - this is just wild. I&#x27;ve seen lots of arguments around &quot;AI won&#x27;t take everyone&#x27;s job, it will just open up new areas for new jobs.&quot; Even if you take that with the benefit of the doubt (which I don&#x27;t really think is warranted):<p>1. You don&#x27;t need to take everyone&#x27;s job. You just need to take a shitload of people&#x27;s jobs. I think a lot of our current sociological problems, problems associated with wealth inequality, etc., are due to the fact that lots of people no longer have competitive enough skills because technology made them obsolete.<p>2. The state of AI progress makes it impossible for humans in many fields to keep up. Imagine if you spent your entire career working on NLP, and now find GPT-4 will run rings around whatever you&#x27;ve done. What do you do now?<p>I mean, does anyone think that things like human translators, medical transcriptionists, court reporters, etc. will exist as jobs at all in 10-20 years? Maybe 1-2 years? It&#x27;s fine to say &quot;great, that can free up people for other thing&quot;, but given our current economic systems, how are these people supposed to eat?<p>EDIT: I see a lot of responses along the lines of &quot;Have you seen the bugs Google&#x2F;Bing Translate has?&quot; or &quot;Imagine how frustrated you get with automated chat bots now!&quot; Gang, the <i>whole point</i> is that GPT-4 blows these existing models out of the water. People <i>who work in these fields</i> are blown away by the huge advances in quality of output in just a short time. So I&#x27;m a bit baffled why folks are comparing the annoyances of ordering at a McDonald&#x27;s automated kiosk to what state-of-the-art LLMs can do. And reminder that the first LLM was only created in 2018.

When I was studying Computational Linguistics I kept running into the unspoken question: given that Google Translate already exists, what is even the point of all of this? We were learning all these ideas about how to model natural language and tag parts of speech using linguistic theory so we could eventually discover that utopian solution that would let us feed two language models into a machine to make it perfectly translate a sentence from one language into another. And here was Google Translate being &quot;good enough&quot; for 80% of all use cases using a &quot;dumb&quot; statistic model that didn&#x27;t even have a coherent concept of what a language is.<p>It&#x27;s been close to two decades and I still wonder if that &quot;pure&quot; approach has any chance of ever turning into something useful. Except now it&#x27;s not just language but &quot;AI&quot; in general: ChatGPT is not an AGI, it&#x27;s a model fed with prose that can generate coherent responses for a given input. It doesn&#x27;t always work out right and it &quot;hallucinates&quot; (i.e. bullshits) more than we&#x27;d like but it feels like this is a more economically viable shot at most use cases for AGI than doing it &quot;right&quot; and attempting to create an actual AGI.<p>We didn&#x27;t need to teach computers how language works in order to get them to provide adequate translations. Maybe we also don&#x27;t need to teach them how the world works in order to get them to provide answers about it. But it will always be a 80% solution because it&#x27;s an evolutionary dead end: it can&#x27;t know things, we have only figured out how to trick it into pretending that it does.

The PR folks at my current company are in full panic mode on Linkedin, judging from the passive-aggressive tone of their posts (sometimes very nearly begging customers not to use ChatGPT and friends).<p>They fully understand that LLMs are stealing lunch money from established information retrieval industry players selling overpriced search algorithms. For a long time, my company was deluded about being protected by insurmountable moats. I&#x27;m watching our PR folks going through the five stages of grief very loudly and very publicly on social media (particularly noticeable on Linkedin).<p>Here&#x27;s a new trend happening these days. Upon releasing new non-fiction books to the general public, authors are simultaneously offering an LLM-based chatbot box where you can ask the book any question.<p>There is no good reason this should not work everywhere else, in exactly the same way. Take for example a large retailer who has a large internal knowledge base. Train an LLM on that corpus, ask the knowledge base any question. And retail is a key target market of my company.<p>Needless to say I&#x27;m looking for employment elsewhere.

I worked in a research capacity in the voice assistant org of a big tech company until very recently. There was a lot of panic when ChatGPT came out, as it became clear that the vast bulk of the org&#x27;s modeling work and research essentially had no future. I feel bad for some of my colleagues who were really specialized in specific NLP technology niches (e.g. building NLU ontologies) which have been made totally obsolete by these generalized LLMs.<p>Personally - I&#x27;m moving to more of a focus on analytical modeling. There is really nothing interesting about deep learning to me anymore. The reality is that any new useful DL models will be coming out of mega-teams in a few companies, where improving output through detailed understanding of modeling is less cost effective than simply increasing data quality and scale. Its all very boring to me.

During my master&#x27;s degree in data science, we had several companies visit our faculty to recruit students. Not a single one was a specialized NLP company, but many of them had NLP projects going on.<p>Most of those projects were the usual &quot;solution looking for a problem to solve&quot;. Even those projects that might have had _some_ utility, would have been way more effective to buy&#x2F;license a product than to develop an in-house solution. Because really, what&#x27;s the use of throwing a dozen 25-30 years old with non-specialized knowledge, when there are companies full of guys with PhDs in NLP that devote all their resources to NLP? Yeah, you can pipe together some python, but these kind of products will always be subpar and more expensive long-term than just buying a proper solution from a specialized company.<p>To me it was pretty clear that those projects were just PR so that c-levels could sell how they were preparing their company for a digital world. Can&#x27;t say I&#x27;m sorry for all the people working on those non-issues though. From the attitude of recruiters and employees, you&#x27;d think they were about to find a cure for cancer. Honestly, I can&#x27;t wait for GPT and other productivity tools to wrech havock upon the tech labour market. Some people in tech really need to be taken down a notch or two.

As one of the comments on reddit posts - it&#x27;s not just big tech companies, but also entire university teams which feel the goalposts moving miles ahead all of a sudden. Imagine working on your PhD on chat bots since start of 2022. Your entire PhD topic might be irrelevant already...

Maybe this is alarmist, but I don&#x27;t see how LLMs don&#x27;t collapse our entire economic system over the next decade or so. This is coming for all of us, not just the NLP experts in big company research groups. Being able to cheaply&#x2F;instantly perform virtually any task is great until you realize there is now nobody left to buy your product or service because the entire middle class has been put out of work by LLMs. And the service industries that depend on those middle class knowledge workers will be out of work because nobody can afford to purchase their services. I don&#x27;t see how this doesn&#x27;t end with guillotines coming out for the owner class and&#x2F;or terrorism against the companies powering this revolution. I hope I&#x27;m wrong.

I tried translating something from English to German (my native language) yesterday with ChatGPT4 and compared it to Microsoft Translate, Google Translate and DeepL.<p>My ranking:<p>1. ChatGPT4 - flawless translation. I was blown away<p>2. DeepL - very close, but one mistake<p>3. Google Translate - good translation, some mistakes<p>4. Microsoft Translate - bad translation, many mistakes<p>I can understand the panic.

They may panic, but they shouldn&#x27;t. They can quickly pivot. GPT programs can be used off the shelf, but they can also use custom training. Every large org has a huge internal set of documents, plus a large external set of documents relevant to its work (research articles, media articles, domain relevant rules and regulations). They can train a GPT bot to their particular codebase. And that is now. Soon (I&#x27;d give it at most one year), we&#x27;ll be able to train GPT bots to videos.<p>All this training does not happen by itself.

Not big tech (or PhD level research), but half the work I did on my side project (subtitles for Chinese learning&#x2F;OCR) is sort of obsolete now, most of the rest of it within a year or two. I put months into an NLP pipeline to segment Chinese sentences, classifying pinyin and translating words in-context, something ChatGPT is great at out the box. My painstaking heuristic for determining show difficulty using word frequencies and comparing distributions to children&#x27;s shows is now the simple task of giving part of the transcript and asking ChatGPT how difficult it is. Next up, the OCR I did will probably be solved by ChatGPT4. It seems the writing is on the wall: most tasks on standard media (text&#x2F;images&#x2F;video), will be &quot;good enough&quot; for non-critical use. The only remaining advantage of bespoke solutions is speed and cost and that will also be a fleeting advantage.<p>But it&#x27;s also extremely exciting, we&#x27;ll be able to build really great things very easily, and focus our efforts elsewhere. Today anyone can throw together a language learning tutor to rival Duolingo. As long as you&#x27;re in it for solving problems you shouldn&#x27;t be too threatened by whatever tool set you&#x27;re currently becoming obsolete.

Everyone here is saying that people can simply transition easily into startups and other big companies. To a certain extent that&#x27;s true, but what exactly are they going to do? As technology consolidates into one or two major LLM&#x27;s, likely only accessible by API, I feel most orgs would be better served by relying heavily on finetuning or optimizing those for their purpose. Previous experience with NLP certainly helps with that, although this type of work would not necessarily be as exciting as trying to build the next big thing, which everyone was scrambling for before.<p>OpenAI could build a state-of-the-art tool with a few hundred developers - to me, that means that money will converge to them and other big orgs rather than the opposite.

I guess I&#x27;m not panicked about my job in the face of AI because <i>objective correctness</i> is required. I <i>dream</i> about the day that OpenAI can write the 100 lines of code that connect the BLE stack, the ADC sensor and the power management code so that my IoT sensor doesn&#x27;t crash once every 8 days.<p>I see the AI stuff as <i>very</i> different from, say, the microcomputer revolution. People had <i>LOTS</i> of things they wanted to use computers for, but the computers were simply too expensive.<p>As soon as microprocessors arrived, people had <i>LOTS</i> of things they were already waiting to apply them to. Factory automation was <i>screaming</i> for computers. Payroll management was <i>screaming</i> for computers.<p>I don&#x27;t see that with the current AI stuff. What thing was waiting for NLP&#x2F;OpenAI to get good enough?<p>Yes, things like computer games opened up whole new vistas, and maybe AI will do that, but that&#x27;s a 20 year later thing. What stuff was screaming for AI right now? Maybe transcription?<p>When I see the search bar on any of my favorite forums suddenly become useful, I&#x27;ll believe that OpenAI stuff actually works.<p>Finally, the real problem is that OpenAI needs to cough up what I want but then it needs to cough up the <i>original references</i> to what I want. I normally don&#x27;t make other humans do that. If I&#x27;m asking someone for advice, I&#x27;ve already ascertained that I can trust them and I&#x27;m probably going to accept their answers. If it&#x27;s random conversation and interesting or unusual, I&#x27;ll mark it, but I&#x27;m not going to incorporate it until I verify.<p>Although, given the current political environment, pehaps I <i>should</i> ask other humans to give me more references.

I&#x27;m not at a big tech company, and we don&#x27;t sell algorithms, but my team does use a lot of NLP stuff in internal algorithms. The only panic I have is trying to keep up and take the time to learn the new stuff. If anything, things like GPT-4 are going to make my team 10x more successful without having to hire an army of PhDs.

I work at a small company, but it’s hard for me to imagine that generative AI will replace predictive AI&#x2F;ML any time soon.<p>Smaller models trained supervised&#x2F;in-domain are simply more efficient and more accurate than unsupervised&#x2F;out-of-domain. Plus we own and operate the technology much more cheaply.<p>I don’t doubt that if your were trying to build a competing product to what OpenAI is doing that you’d feel affected, but there’s also a lot of other problems that are not being solved by generative models.

I think education goal for people shifted. I teach my kids to be flexible and embrace the change. Invest in abilities that transfer well to various things you could be doing during your life. Be a problem solver.<p>In the future -- forget about cosy job you can be doing for the rest of your life. You no longer have any guarantees even if you own the business and even if you are farmer.<p>What you absolutely don&#x27;t want is spend X years at uni learning something, and then 5-10 years into your &quot;career&quot; finding out it was obsoleted overnight and you now don&#x27;t have plan B.

Not even experts in the domain could see themselves being replaced and pivot in time. What hope does an ordinary person have in preparing for what’s coming? Telling people to retrain will not be an acceptable answer because no one can predict which skills will be safe from AI in 5 years.

Fascinating -- I think the comments on the HN post are almost as good.<p>I think everyone mostly agrees that AI is coming for a <i>lot</i> of jobs. There&#x27;s disagreement about how many, how it will impact society and the like.<p>The pace of technology is not linear, it accelerates. I&#x27;ve never seen something that has so rapidly crossed into the &quot;magical&quot; territory as &quot;nearly every single big LLM&#x2F;Generative AI thing&quot; seems to. It redefines what was previously laughably impossible ... a decade ago.<p>We&#x27;re riding a curve upward that is making it extremely hard to see what&#x27;s coming next. All of the pontificating, all of the attempts at finding solutions to imagined problems ... I can&#x27;t see one that doesn&#x27;t feel like a blindfolded person aiming at what they were told was a dart board with what they were told was a dart. There&#x27;s really nothing to do but hang on and hope you land where any new opportunities creep up.<p>Expect bubbles, black swans, and purple unicorns.

Here&#x27;s my two cents, as I work with NLP in a tech company, mostly with Dialogflow and Rasa, in my current project, we are using Chat GPT (and previously GPT 3) to summarize articles, and I see that it can really handle FAQ questions really well. One of most common requirements was to train our bot to handle FAQ type question apart from complex conversation stories&#x2F;flows,but this thing can straight up take the content from an article,summarize it neatly and send a response back.<p>We have had some issues and complaints with the API,( mostly with GPT 3 as the fine tuning was only open for the base model and that had some trouble with some questions). Also there is a finicky response time, despite using having paid access. Response time varies from 10 seconds to even a minute (during some downtime that occured a few days ago, and a few days even before that there was a complete outage).

A whole thread on AI experts discussing how AI is making them obsolete...back to gardening...

If you were an NLP researcher at a university whose past years of experience is facing existential threat due to this rapid innovation causing your area to become obsolete, what would be some good areas to pivot to or refocus on?

It does seem like the (misnamed because it’s not open) OpenAI is very far ahead of most other efforts, especially at the edges in areas like instruction training and output filtering.<p>Playing with Llama 65G gave me a sense for what the median raw effort is probably like. It seems to take a lot of work to fine tune and harness these systems and get them reliably producing useful output.

Some big tech companies are witnessing a panic inside their entire org because they focus almost entirely on their competitors (except for the business divisions which are monopolies).

This is imo a wake-up call about the value of having &quot;AI teams&quot; embedded in companies.<p>Bad analogy- if you had an integrated circuit team in your product company building custom CPUs and Intel came out with the 8080 (or whatever was the first modern commercial chip), probably time to disband the org and use the commercial tech

I would have never guessed that menial jobs would be the hardest ones to automate. I realized that humans have some incredible skills that were vital to our survival, and were honed over billions of years. But we take these skills for granted because most humans are born with these natural abilities, which are actually very impressive in the grand scheme of things. A janitor might not get paid very much relative to other jobs, but it took billions of years of evolution to hone their brain, muscles, balance, hand eye coordination, energy efficiency, information processing, etc. We&#x27;re actually very impressive creatures!

I wonder if this will be a repeat of what happened with speech recognition. It used to be a specialized field dominated by smaller companies like Nuance.<p>More recently Google, Microsoft, Apple, etc. decided they wanted to have speech recognition as an internal piece of their platforms.<p>Google poached lots of Nuance&#x27;s talent. And then Microsoft bought what remained of the company.<p>Now speech recognition is a service integrated into the larger tech company&#x27;s platforms, and also uses their more statistical&#x2F;ML approaches, rather than being a component created by specialist companies&#x2F;groups.<p>(I&#x27;m sure I&#x27;m grossly simplifying this — just seeing a potential parallel.)

As someone who&#x27;s a crap NLP practitioner, everything is just fine and dandy.<p>I&#x27;ve never really had the gear or the skills to put together anything that improves over what I can pull from huggingface.<p>What I do have, and virtually none of my (not remotely technical) colleagues have, is a clue what to do with all this stuff.<p>They reckon it&#x27;s about churning out poems and boilerplate text, the minute I figured it could give me whatever json I could reasonably ask for from a source doc, I was overjoyed.<p>I see more things I can be doing now, not a risk of being replaced.

I see chatGPT as eliminating an awful lot of drudgery.<p>For example, back in the 60&#x27;s my dad was working on his book. The text was typed out double spaced, and he (and others) would make corrections. After a while, my mom would retype the whole thing.<p>Imagine typing a whole book. Again and again and again. She&#x27;d type hour after hour. It&#x27;s dehumanizing.<p>And then came word processors. What a magical revolution! You could edit text instead of typing it all over again. I bet few people today realize what a great achievement that was.<p>All chatgpt does is select the most likely next word out of a corpus of existing text. It is not creative.<p>We don&#x27;t need rooms full of typists anymore. Good riddance. I bet we get rid of a bunch of drudgery jobs with chatgpt.

Maybe semantics, but... The implication of this post is that ChatGPT is causing the demise of NLP jobs, which, sure, that&#x27;s true - but I don&#x27;t think it&#x27;s fair to classify this as replacing the job of anyone who works on NLP. This is not the same as GPT replacing a human translator, which is a direct replacement of their role - it&#x27;s not like ChatGPT is going to start researching NLP. It&#x27;s just a fairly mundane example of a superior technology outpacing an inferior one. The technology of NLP is being obsoleted, not the job of researching NLP (which is just unnecessary, not being replaced with something else). And in fact, NLP researchers are probably some of the most well positioned people to move laterally to working on transformers and LLMs.

My university professor who specialises in NLP kinda feels like what&#x27;s the point of research in the time of chatgpt. He says for now it&#x27;s not possible to scale retrieval easily when using these llms so that&#x27;s what he is looking into for now

Is the entire field of data science (Itself maybe a decade old in terms of being a college major?) now obsolete, in terms of being a distinct job field? Are all data science majors now going to be &quot;just&quot; coming up with the proper prompts to get GPT to correctly massage datasets?

If someone build a software that would be able to attend a meeting instead of you - it will become as popular as Zoom. Combine LLM with voice and video recognition and synthesis, and voila - an immense chunk of wasted time in every corporation is gone.

Is the compute for running an LLM cheap enough to scale at the moment? LLMs seem to be a great generalist solution but could specifically targeted NLP solutions still outperform in terms of speed&#x2F;cost when you are processing high volumes of inputs?

Somebody in that thread is speculating that the company in question is grammarly.<p>I&#x27;m not sure but I&#x27;m now curious as to what the execs there are thinking, especially now with the recent Microsoft 365 news. Feels like the body blows keep coming.

This mostly makes me wonder about the varying definitions of NLP. Most business folks I’ve worked with in this field don’t even know the question to ask or prompt- the implementation details of models like GPT are the easy part.

I look at it the other way. Much of what people use GPT-4 can be done better (with a little more work) with transformer models that are specialized with the task.<p>Things that were a struggle 5 years ago are about to be <i>easy.</i>

We thought AI will take out the mechanical and manual jobs first that don&#x27;t need intellectual capabilities, but only repetition. Who knew AI will first learn to write poems, novels and code.

I remember thinking about this when AlphaFold was announced. Did it happen back then? Were there large shifts in companies&#x2F;universities that were doing folding research?

The book &quot;Who Moved My Cheese?&quot; tries to explain the human behavior for the moments like this. It is happening for many people now.

Sunk cost! This is why so many AI scientists are so skeptical. They don’t use chatGPT because they spent years developing skills that enabled them to use LLMs and now any kid on the street can.

oh no?!<p>so finally the tech sector is experiencing themselves what they have done to other lines of professions for the past decades, namely eradicting them (rightfully) with innovation?<p>well same advice applies then:<p>* embrace, move on and retrain for another profession * learn empathy from the panic and hurt