ASU said today that it will be working with OpenAI to make ChatGPT, the AI-powered chatbot, available to researchers, employees, and faculty at the university.
ASU will launch an open competition in February, inviting staff and faculty to submit ideas for using ChatGPT with a student success focus. fresh directions for investigation and streamlined internal procedures.
Lev Gonick, the chief information officer at ASU, told TechCrunch via email that "as OpenAI begins to explore how to market or align a business for universities, they're looking for a design partner, a thought partner, and I think that's a significant measure of why we've come together today to make this announcement." "ASU will be able to actively participate in defining new uses of artificial intelligence in higher education thanks to our collaboration with OpenAI."
With technology advancing faster than curricula, attitudes towards artificial intelligence in education are changing, as seen by the OpenAI-ASU partnership.
Schools and universities hurried to outlaw ChatGPT last summer because to concerns about plagiarism and false information. Since then, some have lifted their prohibitions, while others have started holding seminars on the educational potential of generative AI technologies.
Unsurprisingly, ASU is fully in favour of using AI as a teaching tool.
"The broad invitation extended to academics, employees, and researchers guarantees that we can offer continuous and significant assistance that establishes the groundwork for our groups to fully utilise these technologies to enhance human education and inventiveness, rather than supplant it," Gonick stated. This strategy is a component of our larger objective to set an example for our students and make sure that the university's use of AI grows in a responsible and scalable way. Our goal is to create an atmosphere where AI resources are applied morally and successfully, laying the groundwork for a more widespread adoption of these tools within our academic community.
In specifics, ASU will give its full-time staff members access to ChatGPT Enterprise accounts. ChatGPT Enterprise was introduced in August and is capable of carrying out the same tasks as ChatGPT, including sending emails, debugging computer code, and creating essays. However, it also enhances performance and customisation possibilities, and adds data analysis and privacy features to the basic ChatGPT.
As an illustration, ChatGPT Enterprise provides an admin panel with capabilities to control ChatGPT usage within an organisation. Users can use ChatGPT to develop internal workflows by using shareable conversation templates, and organisations can create unique ChatGPT-powered solutions by using credits for OpenAI's API platform.
Additionally, ChatGPT Enterprise includes unrestricted access to Advanced Data Analysis, a ChatGPT tool that enables users to have the AI analyse data, make charts, answer math problems, and more, including from uploaded files. Priority access to ChatGPT is also included.
According to Gonick, ASU will manage its ChatGPT Enterprise accounts and work on projects related to the "effective use" and support of AI in a "dual role."
"Our goal in this first phase is to give priority access to our AI tools to ASU faculty, staff, and researchers," he continued. "At the moment, our primary goal is to give our knowledge core control over ChatGPT Enterprise so they can lead the way in both discovery and implementation."
AI
News reports on the firms developing AI and machine learning technologies, the ethical questions they present, and artificial intelligence itself. This includes speech recognition and production, predictive analytics, huge language models, text-to-image and text-to-video models, and generative AI.
Open source vector database startup Qdrant raises $28M
The open source vector database startup Qdrant has raised $28 million in a Series A fundraising round headed by Spark Capital.
The Berlin-based company Qdrant was founded in 2021 and is aiming to capitalise on the rapidly developing AI industry by providing developers with an open source vector search engine and database. This is important because generative AI requires the creation of relationships between unstructured data, which includes text, images, and audio that aren't labelled or otherwise organised, even when the data is "dynamic" and used in real-time applications. According to research from Gartner, unstructured data is expanding three times faster than structured data and accounts for over 90% of all new enterprise data.
The world of vector databases is booming. Weaviate, for example, has raised $50 million for its open source vector database in recent months, and Zilliz has raised $60 million to commercialise the Milvus open source vector database. In another instance, Pinecone obtained $100 million for a proprietary alternative, while Chroma obtained $18 million in venture funding for a project akin to this one.
In April of last year, Qdrant, for example, raised $7.5 million, indicating both a projected expansion by Qdrant and the seemingly endless demand for vector databases among investors.
Qdrant CEO and co-founder Andre Zayarni told TechCrunch, "We received an offer a few months earlier and decided to save some time and start scaling the company now. The plan was to go into the next fundraising in the second quarter this year." "The right people need to be hired and fundraising takes time."
Notably, Zayarni claims that the business turned down a follow-up funding offer at the same time as it was getting a possible takeover offer from a "major database market player." "We went with the investment," he stated, adding that as the company currently primarily consists of engineers, they will utilise the additional funding to expand their commercial staff.
Binary logic
Nine months have passed since Qdrant's last funding round. During that time, the company has introduced binary quantization (BQ), a new, extremely effective compression technology aimed at low-latency, high-throughput indexing that can, according to the company, cut memory usage by up to 32 times and increase retrieval speeds by about 40 times.
According to Zayarni, binary quantization is a technique for "compressing" vectors into the most basic representation using just zeros and ones. The simplest CPU command is to compare the vectors, which greatly accelerates query processing and reduces memory use. Although the theoretical idea is not novel, we applied it in a way that minimises accuracy loss.
however it's up to the user to determine which compression method would work best for their use-cases, BQ might not work for all AI models. According to Zayarni, they discovered that OpenAI's models produced the greatest results, however Cohere and Google's Gemini also performed well. At the moment, Mistral and Stability AI models are being benchmarked against by the company.
These initiatives have drawn prominent adopters including Accenture, Deloitte, and X (formerly Twitter), who is likely the most well-known of them all. Or, perhaps more precisely, Elon Musk's xAI, a business creating Grok, a ChatGPT rival that made its X platform debut last month.
Because of a non-disclosure agreement (NDA), Zayarni was unable to discuss specifics about how X or xAI was using Qdrant; however, it is plausible to conclude that Qdrant was being used to process real-time data. In fact, Grok leverages a generative AI model called Grok-1 that was trained on web data and human feedback. Because of its (current) close alignment with X, Grok can incorporate real-time data from social media posts into its responses; this is known as retrieval augmented generation (RAG), and over the past few months, Elon Musk has publicly hinted at use-cases for this technology.
Qdrant does not disclose which of its clients are using its managed services and which are using the open source Qdrant incarnation, but it did identify several startups that are "mostly" utilising its managed cloud service, including GitBook, VoiceFlow, and Dust. This effectively relieves resource-constrained businesses of the burden of managing and deploying everything on their own, as they would have to do with the core open source incarnation.
Even if a business chooses to pay for additional services, Zayarni is certain that one of the main selling elements is the company's open source credentials.
"There is always a risk of vendor lock-in when using a proprietary or cloud-only solution," Zayarni stated. Customers must consent to price adjustments or other changes made by the vendor, or they will have to think about switching to an other solution, which will be difficult in the case of a heavy-production use case. You have constant control over your data while using open source, and you may choose from a variety of deployment choices.
In addition to announcing the fundraising today, Qdrant is also formally launching its managed "on-premise" edition, which allows businesses to host everything in-house while still utilising Qdrant's premium features and support. This comes after it was announced last week that Qdrand's cloud edition would be available on Microsoft Azure in addition to the platforms already supported by AWS and Google Cloud.
Unusual Ventures and 42cap participated in Qdrant's Series A financing in addition to lead investor Spark Capitali.
It’s 2021 for AI while the rest of the startup market is stuck in 2024
Hi there, and welcome back to Equity, the podcast where we analyse the data and nuances that lie beneath the headlines about the business of startups.
We review the previous weekend and look ahead to the upcoming week on this Monday podcast. This time around, there was so much news that we had to narrow down our discussion topics significantly. tiny mercies!
Here’s the rundown:
- This week marks the start of the Q4 2023 earnings cycle, with major companies like Visa and Intel releasing their results. Next week is when the biggest tech businesses will begin to report.
- For reasons that are still unclear, cryptocurrency is not seeing a post-ETF boom. However, when have fluctuations in cryptocurrency prices ever made total sense?
- ElevenLabs is the most recent unicorn in AI. The synthetic voice business now has plenty of cash to attempt to manage its market thanks to a recent $80 million investment.
- Canva is big! Who knew?
- Crunchbase News also states that last year saw a decline in cybersecurity fundraising. That is strange, considering the amount of market breaches, don't you think?
- There is a disagreement between Apple Vision Pro and developers. Could Apple try to mend its ties with developers?
- Finally, it appears that TikTok can be slowed down.
Like I said, it’s a busy start to the week! Talk soon!
Visit Equity's Simplecast website for more information, including transcripts of the episodes.
Every Monday, Wednesday, and Friday at 7 a.m. PT, Equity drops, so make sure to follow us on Apple Podcasts, Overcast, Spotify, and all the casts. A fantastic programme on cryptocurrency that features founder interviews and more can be seen on TechCrunch!
Turn headwinds into opportunity in 2024
The founders I work with are aware of how often I think of John Coltrane. I've been reflecting lately on how he revolutionised jazz by utilising what are known as "Coltrane changes," which are harmonic progressions.
Coltrane transitions, made famous by his record "Giant Steps" in 1960, are distinguished by quick and frequent modulations between key centres. The intricate progressions broke the mould of conventional jazz improvisation, forcing players to experiment with new scales and rhythms in order to adjust to the changes. They had an impact on how jazz developed into what it is today.
How is any of this relevant to launching a business? A lot, in a year like 2023.
In terms of business, 2023 was a year when organisations had to revert to the fundamentals and modify their plans in response to a macroeconomic climate that was prone to volatility.
That required the founders to reconsider how they were creating and expanding. It meant viewing cash as the necessary resource to be alive and present as a static item on the balance sheet. It required making difficult decisions about hiring, carefully considering who was essential, and putting knowledge ahead of loyalty. It meant going above and beyond to secure their product's position as a need rather than a nice-to-have in a market that was still apprehensive about the full impact of artificial intelligence.
Additionally, it was an exceptional year for investors. One could recall the AI craze of the past, when everyone was trying to found the next big AI startup. However, a lot of would-be business owners chose to stay out of the game, either because they were burnt by cryptocurrency or because they believed funding would be too challenging.
In my interactions with founders, I've made an effort to be a voice of reason. Being flexible is crucial, and startups are more like marathons than sprints. We may conclude from historical downturns that some of the greatest businesses and leaders have emerged from them. Similar to this, "Giant Steps" pushed performers to be creative in order to keep up with Coltrane's quick changes.
It is now, in 2024, that entrepreneurs need to unleash their creativity and develop the fortitude, acumen, and self-control that will see them through the next two decades.
Get ready for the next wave of generational startups
Throughout history, we have witnessed the greatest entrepreneurs stepping up when it is difficult to raise capital during economic downturns.
Risk-taking is the essence of entrepreneurship; it entails chasing audacious ideas, venturing into the unknown, and creating without fear of failure.
Consider some of the most creative and prosperous businesses of the last 20 years: many of the well-known brands you see today, including Square, Uber, Airbnb, and Stripe, were founded in the wake of the 2008 financial crisis. These businesses, led by visionary entrepreneurs, acted with a focus, discipline, and entrepreneurial spirit that makes them superpowers in times of scarcity. They grabbed hold of concepts that they thought could upend established markets and industries.
When Dropbox obtained its Series A funding round in 2008, the company employed nine people. In addition to having a clear idea of how cloud storage would change file storage and collaboration, Drew Houston also brought a scarcity mentality to the organisation, which made resource allocation more innovative and effective. Even though the company had only added a small number of staff, Dropbox had more than 45 million users by the time we led its Series B in 2011.
When Dropbox obtained its Series A funding round in 2008, the company employed nine people. In addition to having a clear idea of how cloud storage would change file storage and collaboration, Drew Houston also brought a scarcity mentality to the organisation, which made resource allocation more innovative and effective. Even though the company had only added a small number of staff, Dropbox had more than 45 million users by the time we led its Series B in 2011.
AI will be at the forefront of that wave — led by visionary entrepreneurs
2024 will see AI continue to rule the news. But what really interests me is watching how AI technology is developed and marketed, as well as how business owners plan to use it in regular operations.
It can be challenging to distinguish between the hoopla and the real-world applications of AI since ChatGPT startled the world a year ago. However, the dust is beginning to clear, and new businesses are emerging with a true entrepreneurial focus on how AI can be used to develop pertinent goods and services.
In 2024, this trend will only pick up speed as every business formulates an AI strategy and starts integrating AI into its daily operations. This paradigm change will pave the way for a fresh round of market disruption, lifting artificial intelligence (AI) above the hype and positioning it as the cornerstone of the following wave of truly inventive firms.
I'm especially curious to watch how the upcoming generation of driven businesspeople will seize this chance. Recall that university academics played a major role in driving innovation in the early days of artificial intelligence. These organisations have played a crucial part in getting us to where we are now and will continue to do so as technology advances at a breakneck speed. However, there is a distinction between developing a product that adds value for a specific market and inventing in a lab to solve a challenging technical problem.
We were so impressed with the founders' approach to productization that we decided to invest in Cohere two years ago. Aidan, Ivan, and Nick were true researchers who had studied under esteemed scholars such as Geoffrey Hinton, dubbed "the godfather of AI." However, they also had a distinct idea about how to turn huge language models into products that might assist enterprise businesses in creating useful, day-to-day business applications.
When we oversaw the seed and Series A rounds of biotech firm Cradle, we had the similar feelings. Stef and his co-founders have not only discovered a strong need for their product among R&D teams, with massive upside given the market scale, but they also possess a unique combination of deep machine learning skills and protein engineering experience from top tech and biotech organisations.
The development of AI is still in its infancy. AI will need time to develop into its final form, much like Yahoo paved the way for Google or MySpace for Facebook. Visionary founders are currently preparing to launch the next generation of generational enterprises by researching and learning from advances in artificial intelligence.
Dormant sectors are in for an AI awakening
Being able to view particular industries in a new perspective because of the promise of artificial intelligence was one of my favourite and most unexpected lessons from 2023. That will only pick up speed going forwards.
Marketing is a prime illustration of this. We haven't seen any significant advancements in advertising technology in a long. Nevertheless, I believe significant changes in that sector are imminent given the increasing accessibility and sophistication of targeting and personalisation made possible by AI, as well as the mostly unrealized promise of predictive analytics and programmatic advertising.
Another industry that could use some fresh disruption is dating. Dating is a very intimate human experience, as we all know. Although it has made connections easier, online dating has also presented new difficulties. Some would contend that incorporating AI will make dating applications less human. However, I envision the opposite: these apps may enable individuals to concentrate more on interpersonal relationships through improved matching algorithms, more tailored suggestions, enhanced security, or even features that leverage augmented or virtual reality. In this industry, there's a chance for the person who can find the appropriate balance to lead.
And then there are all the other industries that I have always been enthusiastic about and believe are ripe for innovation: personal productivity apps, gaming, and the creator class. I'm excited to watch new leaders emerge and see how AI elevates these industries in 2024.
Regulating AI will be a global responsibility
Being the complete opposite of a nationalist, I find it odd when nation-first rhetoric permeates startup culture.
Artificial intelligence (AI) is a tremendously transformative technology with genuine concerns that are beginning to surface. Naturally, we must exercise caution in its application, but discussing these difficult problems in nationalistic terms serves to divert attention from the main goal, which is to make sure that these technologies are used in a morally and securely manner. International cooperation will be necessary to get this right.
Keep in mind that the majority of AI technologies are cross-border in nature; the businesses that create and implement them are international, thus this means that their influence is felt beyond legal boundaries. Variations in national approaches will cause fragmentation and inconsistency between nations, exposing weaknesses, stifling innovation, and producing a patchwork of laws that aren't as good as they could be.
Geopolitical divides may make international regulation more difficult and complex, but a global strategy is necessary to provide sufficient barriers to AI's morally and safely used applications and to guarantee an environment that fosters AI innovation. The focus of the discussion needs to change from controlling fundamental technology on the basis of a fictitious AI apocalyptic threat to addressing the real-world use cases and developing risks.
What is the best way for founders to think about converting obstacles into opportunities? The most successful businesspeople manage to block out the distractions and carry out their mission as only they can.
The days of “low-risk, high-reward” are gone
A generation of entrepreneurs has been duped into thinking that enormous profits are achievable without risk, that you can ride a magic carpet made of money to the top of the mountain, all thanks to historically low interest rates. I apologise, but that was an illusion.
The essence of entrepreneurship is taking chances. And by real, revolutionary risk, I do not mean incremental risk. This entails chasing bold ideas, venturing into the unknown, and innovating fearlessly. It entails placing wagers with a growth mentality, overcoming setbacks with fortitude, and having the audacity to keep attempting things that aren't sure to succeed.
Stewart Butterfield, one of the co-founders of Slack, is the most knowledgeable person about topic. Butterfield has had the conviction to create a massively multiplayer online role-playing game not once, but twice in his career. On both occasions, he had the guts to change course when he realised his experiments were failing. In the first instance, Butterfield sold Flickr to Yahoo just 12 months after it had its formal launch. Flickr had started out as a shareable in-game photo inventory.
A few years later, a similar tale came to pass when Butterfield closed down Glitch, his second game, after realising it would not generate any revenue. After raising $15 million to create Glitch, his firm changed its direction and concentrated on creating an internal communication platform. The rest of the narrative doesn't need to be told: in only two years after going public, Slack had raised $340 million, drawn over 2 million daily active users, and won the 2015 Inc. Company of the Year award. Salesforce paid $27.7 billion to acquire Slack five years later.
Low-risk founders are at a disadvantage to rivals who aren't afraid to take chances and pursue more innovative ideas. As an investor, I will always support the founder who has faith in their idea and isn't afraid to take a risk because that's where the biggest returns are found.
What about failure? Dreaming large will always lead to this. Acquiring knowledge from your mistakes is crucial. Recall the advice of Samuel Beckett: "Try again. Err once more. Perform better when you fail.
Discipline is more important than big valuations
My observation is that a company's success is frequently inversely correlated with the amount of money collected in its initial funding round, and I frequently share this with founders.
Some of the largest success stories in our portfolio firms had modest beginnings, as I can see when I look at them. With a $38 billion market valuation as of right now, Datadog raised $6.2 million in its Series A funding round. Seed financing for Figma totaled $3.9 million at launch. Discord had $1.1 million upon launch. Roblox spent $560,000 on its Series A.
These businesses and their founders are excellent examples of how an early scarcity mindset can remove optionality and distractions and create discipline, which is one of the most critical traits any entrepreneur can have. Discipline is then free to focus on what matters most to the success of the business.
We were instantly captivated by Pieter and Arnout, the founders of Adyen, and their aim of developing a worldwide payments solution when we first met them in 2011. Self-centered? Yes, particularly for a tiny Dutch business operating in a highly regulated sector. However, the business had already turned a profit and had signed up clients on four continents. We had to persuade them to let us lead their Series A because they were so disciplined they didn't need our money.
In 2024, when investing becomes more active, I'm confident we'll witness some astounding valuations. Don't overthink the fact that these high expectations inevitably lead to success. As many successful businesses have their modest beginnings, so too do many companies that have raised large initial rounds yet failed because of internal issues, a lack of discipline, or simply being outperformed by the competition.
Don’t sacrifice growth for profitability at all costs
Profitability is the only factor that matters, according to people on Wall Street. But you cannot manage your company in accordance with Wall Street's wishes. It's the equivalent of letting your tail wag your dog in the business world.
Profitability is important, of course, but you shouldn't sacrifice long-term goals for immediate efficiency. This relates to having a clear vision and being willing to take chances. Businesses that can expand efficiently and profitably while maintaining higher margins are the most successful. The first step in solving that equation is determining how to promote growth.
Nobody would have objected if Figma had held another tiny developer conference in 2023. However, they recognised their chance when everyone was watching them following the now-canceled Adobe deal and no one else was sponsoring or funding significant developer conferences. They organised their largest conference yet, taking a chance. And what do you know? With almost 8,500 attendees, it was a huge success. It fundamentally altered the way the market views Figma and provided them with a tested lever to pull in order to spur even greater growth in the coming years.
It always comes back to basics
Although we are addicted to novelty as humans, newer doesn't always mean better. Not always is bigger better. Furthermore, there is always a market for something, regardless of how novel or fascinating it is.
The world is evolving more quickly than before. There won't be any innovation like this in history in 2024. Although I'm looking forward to it, I'm also careful not to get carried away by the excitement. Regardless of your role as investor or entrepreneur, it's important to keep in mind that the fundamental components of a profitable company have remained constant:
- Visionary leadership.
- A clear value proposition.
- A well-defined market.
- A product or service that provides real value.
We were confident enough to invest in Figma in 2013 because of these values. Dylan was a 19-year-old intern at LinkedIn when we first spoke. On paper, there was no reason for anyone to put money into him and Evan. However, we were convinced by their vision—and, more crucially, by their commitment to become the world's most significant product design firm.
We at Index have always made no secret of the fact that we prioritise investing in people. Establishing a company is an art, and the entrepreneur is the master craftsman. Although we as investors try our best to empower and assist them, the entrepreneur remains the key player and the only one who truly knows what's best for their company.
The businesses that best exemplify the essence of entrepreneurship in 2024—which revolves around having lofty goals, a captivating vision, and unwavering commitment to the cause—will be the most prosperous. I'm eager to witness who steps forward, what their goal entails, and how we can contribute by helping them along the way.
New MIT CSAIL study suggests that AI won’t steal as many jobs as expected
What jobs will AI automate if any, and when will it do so?
A recent study released this morning by MIT's Computer Science and Artificial Intelligence Laboratory (CSAIL) aims to address these three problems.
Many attempts have been made to forecast and extrapolate how current AI technology, such as massive language models, might affect people's livelihoods and entire economies in the future.
According to Goldman Sachs, in the coming years, artificial intelligence could automate 25% of the labour market. By 2055, over half of all labour will be led by AI, predicts McKinsey. According to a poll conducted by Princeton, NYU, and the University of Pennsylvania, ChatGPT alone may have an impact on 80% of occupations. Furthermore, according to a research by the recruitment agency Challenger, Grey & Christmas, AI is already displacing thousands of workers.
However, the goal of the MIT study was to go beyond what they refer to as "task-based" comparisons and evaluate the likelihood that firms will truly replace human labour with AI technology as well as the viability of AI doing certain functions.
In contrast to what one might anticipate, including this writer, the MIT researchers discovered that most tasks that were previously considered vulnerable to AI displacement aren't, at least not yet, "economically beneficial" to automate.
Neil Thompson, a research scientist at MIT CSAIL and a co-author of the paper, believes that the main lesson is that the impending AI disruption may occur more gradually and subtly than some pundits have predicted.
In an email conversation with TechCrunch, Thompson stated, "Like much of the recent research, we find significant potential for AI to automate tasks." However, we can demonstrate that a large number of these tasks are not yet well-suited for automation.
A crucial disclaimer is that the study only included positions involving visual analysis, such as quality-control product inspection at the end of a production line. The researchers left it to future study to examine the possible effects of text- and image-generating models, such as ChatGPT and Midjourney, on labourers and the economy.
The purpose of this study was to determine what tasks an AI system would need to complete in order to completely replace workers, hence the researchers conducted a poll of workers. The cost of developing an AI system that could accomplish all of this was then estimated, and it was also determined whether or not companies, particularly "non-farm" American companies, would be prepared to cover the system's startup and ongoing costs.
Early in the study, the researchers give the example of a baker.
The U.S. Bureau of Labour Statistics estimates that a baker spends roughly 6% of their time ensuring food quality, a duty that artificial intelligence (AI) might (and is) automating. If a bakery with five employees and an annual salary of $48,000 were to automate food quality checks, it might save $14,000. However, based on the study's estimations, a minimal, in-house AI system capable of performing the task would cost, at most, $165,000 to implement and $122,840 annually to operate.
According to Thompson, "we find that automating vision tasks with AI would be economically attractive at only 23 percent of the wages being paid to humans for doing so." "Doing these types of jobs is still a better economic choice for humans."
The report does, however, take into consideration self-hosted, self-service AI systems offered by companies such as OpenAI, which simply require task-specific fine-tuning rather than complete training. However, the researchers claim that even with a system that can be purchased for as little as $1,000, there are many low-wage, multitasking-dependent positions that are too many for a corporation to economically automate.
The researchers note in the paper that "even if we consider the impact of computer vision just within vision tasks, we find that the rate of job loss is lower than that already experienced in the economy." "It would still take decades for computer vision tasks to become economically efficient for firms, even with rapid cost reductions of 20% annually."
The researchers acknowledge a number of limitations with the study, which is commendable. For instance, it ignores situations in which artificial intelligence (AI) can supplement human labour rather than replace it (e.g., analyse a golfer's swing) or develop entirely new roles and responsibilities (e.g., maintain an AI system). Furthermore, it does not account for all potential cost savings from pre-trained models such as GPT-4.
One wonders if the MIT-IBM Watson AI Lab, the study's sponsor, put any pressure on the researchers to come to a particular conclusion. With IBM's $240 million, ten-year donation, the MIT-IBM Watson AI Lab was established. IBM has a stake in making sure that AI is seen as nonthreatening.
But the researchers assert this isn’t the case.
“We were motivated by the enormous success of deep learning, the leading form of AI, across many tasks and the desire to understand what this would mean for the automation of human jobs,” Thompson said. “For policymakers, our results should reinforce the importance of preparing for AI job automation . . . But our results also reveal that this process will take years, or even decades, to unfold and thus that there is time for policy initiatives to be put into place. For AI researchers and developers, this work points to the importance of decreasing the costs of AI deployments and of increasing the scope of how they can be deployed. These will be important for making AI economically attractive for firms to use for automation.”
Voice cloning startup ElevenLabs lands $80M, achieves unicorn status
There’s a lot of money in voice cloning.
As an example, consider ElevenLabs, a firm that is creating artificial intelligence-powered tools for producing and editing synthetic voices. Today, the company announced that it has secured a $80 million Series B investment, which was co-led by well-known investors, such as Andreessen Horowitz, entrepreneur Daniel Gross, and former GitHub CEO Nat Friedman.
With participation from Sequoia Capital, Smash Capital, SV Angel, BroadLight Capital, and Credo Ventures, the round values ElevenLabs at over $1 billion, up from over $100 million in June of last year. ElevenLabs has received $101 million in total. According to CEO Mati Staniszewski, the additional funds will go towards developing new products, growing the staff and infrastructure of ElevenLabs, conducting AI research, and "improving safety measures to ensure responsible and ethical development of AI technology."
In an email interview with Newsreedom, Staniszewski stated, "We raised the new money to cement ElevenLabs' position as the global leader in voice AI research and product deployment."
ElevenLabs, which was co-founded in 2022 by Staniszewski, a former Palantir deployment strategist, and Piotr Dabkowski, an ex-Google machine learning engineer, debuted in beta about a year ago. Staniszewski claims that badly dubbed American movies served as inspiration for him and Dabkowski, both of whom are Polish natives, to develop voice cloning software. They believed AI could perform better.
Currently, ElevenLabs' most well-known product is probably its browser-based speech synthesis application, which produces realistic voices with modifiable toggles for intonation, emotion, cadence, and other essential vocal aspects. Users can submit text and have it read aloud by one of several default voices on a recording for free. ElevenLabs' voice cloning technology allows paying clients to upload voice samples for use in creating new styles.
ElevenLabs is spending more and more in voice-generating technologies to produce audiobooks, dub films and TV series, and create character voices for video games and promotional campaigns.
The business unveiled a "speech to speech" tool last year that aims to maintain a speaker's voice, prosody, and intonation while automatically eliminating background noise. It also translates and synchronises speech with the original content in the case of films and TV shows. A new dubbing studio process that includes tools for creating and editing transcripts and translations, as well as a mobile app that is available for a subscription and narrates text and webpages utilising ElevenLabs voices, are planned for the upcoming weeks.
ElevenLabs' innovations have gained the startup clients of many publishing, media, and entertainment firms, including The Washington Post and Paradox Interactive, a game developer whose recent projects include Cities: Skylines II and Stellaris. Staniszewski asserts that 41% of Fortune 500 businesses' employees use ElevenLab, and that the platform has produced the equivalent of more than a century's worth of audio.
But the publicity hasn’t been totally positive.
Using ElevenLabs' technologies, the notorious message board 4chan—which gained notoriety for its conspiratorial content—shared abusive messages that imitated well-known figures, including actress Emma Watson. In a couple of seconds, James Vincent of The Verge was able to use ElevenLabs to maliciously clone voices, producing samples that included anything from violent threats to derogatory remarks about transgender people. Additionally, reporter Joseph Cox at Vice published footage of creating a clone that was realistic enough to trick an authentication system at a bank.
As a result, ElevenLabs has launched a tool to identify speech produced by its platform and made an effort to remove users who violate its terms of service, which forbid abuse. According to Staniszewski, ElevenLabs intends to work with unidentified "distribution players" to make the tool available on third-party platforms this year and to enhance the detection tool's ability to identify sounds from other voice-generating AI models.
 |
| ElevenLabs offers an array of different voices, some synthetic, some cloned from voice actors |
Voice actors have also criticised ElevenLabs, claiming that the company utilises voice samples of them without permission, which might be used to promote content they don't support or disseminate false information. In a recent Vice piece, victims describe how ElevenLabs was utilised in harassment campaigns against them. One such instance was the use of a cloned voice to reveal an actor's home location, which is considered private information.
Then there's the big issue: ElevenLabs and other platforms pose an existential threat to the voice acting business.
According to Motherboard, voice actors are increasingly being requested to give over the rights to their voices so that companies can utilise artificial intelligence (AI) to create synthetic replicas of them that may one day take their place, sometimes without paying them a fair price. Actors worry that voice work, especially low-paying, entry-level jobs, may soon be supplanted by artificial intelligence-generated voices and that they won't have any other options.
Some platforms are attempting to achieve equilibrium. Replica Studios, an ElevenLabs rival, and SAG-AFTRA inked an agreement earlier this month for the production and licencing of digital voices of media artist union members. The organisations stated in a press statement that the agreement included "fair" and "ethical" terms and conditions to guarantee performer permission, including negotiating arrangements for digital voice duplicates in new works.
However, several voice performers, notably SAG-AFTRA members, weren't happy with even this.
Marketplace for voices is ElevenLabs' answer. The marketplace, which is now in alpha and will roll out more broadly in the coming weeks, enables users to record, authenticate, and distribute their voices. According to Staniszewski, original creators are paid when someone else uses their voice.
He went on, "Users always maintain control over the availability and compensation terms of their voice." "The marketplace brings a diverse range of voices to ElevenLabs' platform and is designed as a step towards harmonising AI advancements with established industry practices."
However, voice performers might not agree with ElevenLabs' current practice of not paying in cash. As things stand right now, ElevenLabs' premium services are credited to creators—a situation that some may find hilarious, I imagine.
Maybe that will change in the future as ElevenLabs, one of the best-funded synthetic voice businesses now operating, tries to defeat Big Tech heavyweights like Amazon, Microsoft, and Google in addition to up-and-coming competitors like Papercup, Deepdub, Acapela, Respeecher, and Voice.ai. Regardless, ElevenLabs hopes to stay in the rapidly expanding synthetic voice business and make ripples in it. The company plans to increase its personnel from 40 to 100 by the end of the year.