Connect with us

Published

on

OpenAI on Monday launched a new AI model and desktop version of ChatGPT, along with an updated user interface, the company’s latest effort to expand use of its popular chatbot.

The update brings GPT-4 to everyone, including OpenAI’s free users, technology chief Mira Murati said in a livestreamed event. She added that the new model, GPT-4o, is “much faster,” with improved capabilities in text, video and audio.

“This is the first time that we are really making a huge step forward when it comes to the ease of use,” Murati said.

OpenAI, backed by Microsoft, has been valued by more than $80 billion by investors. The company, founded in 2015, is under pressure to stay on top of the generative AI market while finding ways to make money as it spends massive sums on processors and infrastructure to build and train its models.

The new model also has improved quality and speed of ChatGPT for 50 different languages, and it will also be available via OpenAI’s API so that developers can begin building applications using the new model today, Murati said. GPT-4o is twice as fast as, and half the cost of, GPT-4 Turbo, Murati said.

OpenAI team members demonstrated the new model’s audio capabilities, asking for help calming down ahead of a public speech. OpenAI researcher Mark Chen said the model has the capability to “perceive your emotion,” adding that the model can also handle users interrupting it. The team also asked it to analyze a user’s facial expression to comment on the emotions the person may be experiencing.

“Hey there, what’s up? How can I brighten your day today?” ChatGPT’s audio mode said when a user greeted it.

Chen demonstrated the model’s ability to tell a bedtime story and asked it to change the tone of its voice to be more dramatic or robotic. He even asked it to sing the story.

OpenAI’s new model can also function as a translator, even in audio mode, the company said. Chen demonstrated the tool’s ability to listen to Murati speaking Italian while he spoke English and to translate into their respective languages as they conversed.

Team members also demonstrated the model’s ability to solve math equations and help write code, positioning it as a stronger competitor to Microsoft’s own GitHub Copilot.

For OpenAI, it’s one of the company’s biggest announcements since its August launch of ChatGPT Enterprise, the AI chatbot’s business tier. That tool was in development for “under a year” and had the help of more than 20 companies of varying sizes and industries, OpenAI COO Brad Lightcap told CNBC at the time.

OpenAI, Microsoft and Google are at the helm of a generative AI gold rush as companies in seemingly every industry race to add AI-powered chatbots and agents to key services to avoid being left behind by competitors. Earlier this month, OpenAI rival Anthropic announced its first-ever enterprise offering and a free iPhone app.

A record $29.1 billion was invested across nearly 700 generative AI deals in 2023, an increase of more than 260% from the prior year, according to PitchBook. The market is predicted to top $1 trillion in revenue within a decade.

Some in the industry have raised concerns about the speed at which untested new services are coming to market, and academics and ethicists are distressed about the technology’s tendency to propagate bias.

After ChatGPT’s launch in November 2022, it broke records at the time as the fastest-growing consumer app in history, and now has about 100 million weekly active users. OpenAI says that more than 92% of Fortune 500 companies are using the platform.

Murati said during the Monday event that OpenAI wants to “remove some of the mysticism from the technology.”

“Over the next few weeks, we’ll be rolling out these capabilities to everyone,” Murati said, adding.

She concluded by thanking Nvidia CEO Jensen Huang and his company for providing the necessary graphics processing units (GPUs) to power OpenAI’s technology.

“I just want to thank the incredible OpenAI team, and also thanks to Jensen and the Nvidia team for bringing us the most advanced GPUs to make this demo possible today,” she said.

WATCH: OpenAI’s Google search competitor

Continue Reading

Technology

Nvidia reports first-quarter earnings after the bell

Published

on

By

Nvidia reports first-quarter earnings after the bell

Jensen Huang, co-founder and chief executive officer of Nvidia Corp., during the Nvidia GPU Technology Conference (GTC) in San Jose, California, US, on Tuesday, March 19, 2024. 

David Paul Morris | Bloomberg | Getty Images

Nvidia reports fiscal first-quarter earnings on Wednesday after the bell.

Here’s what Wall Street expects, per LSEG consensus estimates:

  • Earnings Per Share: $5.59, adjusted
  • Revenue: $24.65 billion

The chipmaker, which a decade ago was a niche developer of 3D gaming hardware, has found itself at the center of the action in technology.

Nvidia’s report comes about a year after the company first signaled to investors that it was about to embark on a stretch of torrid growth powered by demand for artificial intelligence chips from companies such as Google, Microsoft, Meta, Amazon and OpenAI.

Revenue has increased by more than 200% in each of the past two quarters, and Wall Street is expecting that trend to continue, with estimates showing a 243% surge in the first quarter from a year earlier. Net income is expected to be up more than fivefold from a year ago.

Nvidia shares have more than tripled since the company reported fiscal first-quarter earnings last year and provided surprisingly strong guidance for the second quarter.

The company’s current generation of AI graphics processing units (GPUs), called Hopper, are required by the top AI scientists to develop chatbots, translators and image generators. For the past year, customers have been buying them up in droves, with the top cloud and internet companies spending billions of dollars on the technology to build out their infrastructure.

But questions are swirling about the sustainability of Nvidia’s meteoric growth as many customers have to start showing a profit from all their hefty expenditures. AI software costs significantly more to run than traditional software, partially due to the outlay necessary for Nvidia GPUs.

Nvidia is also starting to ship its next-generation AI GPUs, called Blackwell. Some businesses may be eyeing the upcoming chips, causing a possible lull in sales of the existing technology.

Starting in the fiscal second quarter, Nvidia will bump up against tough year-over-year comparisons to the initial days of AI-driven growth. Analysts expect expansion to dip below 100% in the July quarter and decelerate significantly over the following two periods.

Don’t miss these exclusives from CNBC PRO

Options Action: Nvidia options pricing unprecedented market cap move on earnings

Continue Reading

Technology

A Microsoft under attack from government and tech rivals after ‘preventable’ hack ties executive pay to cyberthreats

Published

on

By

A Microsoft under attack from government and tech rivals after 'preventable' hack ties executive pay to cyberthreats

Microsoft has come under fire recently from both the U.S. government and rival companies for its failure to stop a Chinese hack of its systems last summer. One change the tech giant is making in response: linking executive compensation more closely to cybersecurity.

In April, a government review board described a hack of Microsoft last summer attributed to China as “preventable.” The U.S. Department of Homeland Security’s Cyber Safety Review Board pointed to “a cascade of errors” and a corporate culture at Microsoft “that deprioritized enterprise security investments and rigorous risk management.”

Competitors have taken advantage of the cyber lapse, with Google publishing a blog post this week highlighting the government findings and noting, “The CSRB report also highlights how many vendors, including Google, are already doing the right thing by engineering approaches that protect against tactics illustrated in the report.” 

CrowdStrike prominently displays the government conclusions on its site.

Nation-state attacks from China and Russia are increasing, and targeting corporations across the economy, as well as the U.S. government and social infrastructure. Microsoft has been a very big target, including hacks by Russia and China. There is growing pressure from the U.S. government for the company to improve its cybersecurity protocols, with its top corporate lawyer, Brad Smith, being called to testify on Capitol Hill.

Microsoft is in damage control mode. After a hack of executive email accounts in January attributed to Russian hackers, the company disclosed the incident in compliance with new federal cybersecurity disclosure rules, even though technically it was not a “material” hack that it was required by law to share, leading to discussion at other firms about where to draw the line on the new disclosure. The decision by Microsoft to link executive compensation to successful cybersecurity performance is another is prompting discussions at other firms. 

Microsoft launched its Secure Future Initiative in November, and earlier this month, the company outlined in a blog post from Charlie Bell, executive vice president of Microsoft Security, that as part of its SFI goals it will “instill accountability by basing part of the compensation of the company’s Senior Leadership Team on our progress in meeting our security plans and milestones.”

A Microsoft spokesperson declined to provide specifics on the compensation, but said as a company which plays a central role in the world’s digital ecosystem, it has a “critical responsibility” to make cybersecurity a top priority. It is part of the company’s “important governance changes [made] to further support a security-first culture,” the spokesperson said. 

Companies often provide more details, though often only limited details, on executive compensation performance targets in annual meeting proxies, which in Microsoft’s case was last held in December 2023.

Cybersecurity as a core corporate risk and bonus metric

It has become more common for corporations to tie a percentage of annual executive bonus payouts to various goals that go beyond meeting sales and profit targets. In recent years, many Fortune 500 companies, including Apple, have added bonus pay tied to ESG metrics. Risk management and safety goals have long been a part of executive compensation, dating back to an era before the rise of ESG — for example, mining and energy companies, as well as manufacturers and industrials, tying bonuses to environmental and worker safety.

The conversations about cybersecurity-linked executive pay have started taking place at other companies since Microsoft made its move, according to Aalap Shah, managing director at executive compensation consultant Pearl Meyer. It’s not prevalent as a compensation practice today, he said, but he added, “post-Microsoft’s announcement, I’ve gotten phone calls asking, ‘Should we do it? Would it work?’ … These conversations are very similar to the ones we were having a few years ago with ESG metrics and a significant percentage of companies adopted them.”

Shah said there is a case to be made that cybersecurity is a core issue that can be equated to mining or industrial safety. But there’s a big difference between a business in cybersecurity and, for example, a retailer, in making this case. And even in industries beyond technology and cybersecurity where keeping data secure is a core issue, such as financial services and health care — which have been targets of high-profile hacks — it’s not a clear case yet to tie executive compensation of the most senior people, such as a chief financial officer or general counsel, to cybersecurity, versus the chief information security officer or chief technology officer, specifically.

Tying pay to hacks is a ‘good place to start’

Some firms will make the case that cybersecurity is already ingrained in their culture and such a move would be redundant, but with the escalation in hacking threats and increased importance of cybersecurity spending to the bottom line of companies like Microsoft, this new executive pay metric may be overdue.

Making executive compensation contingent, to some degree, on meeting cybersecurity aims is a good place to start instilling a security culture at the top of the corporate hierarchy that is fundamental to success, according to experts. 

“The most important message being sent internally and externally is it’s very important to their culture and more and more companies will follow suit, regardless of whether the gain is significant,” Shah said. “What they want to do is make sure it is becoming ingrained culturally, and the path to do that is by linking it to compensation.”

“Cybersecurity has to be in the culture of the organization,” said Stuart Madnick, professor of information technology at MIT. But prioritizing security can be difficult within a corporation, Madnick said, because it often means putting money into places that aren’t clearly reflected on the bottom line. “Corporate culture prioritizes other things over security and risk management,” Madnick said. “How do you know how secure you are? Maybe no one is targeting you at the time. But if you increase sales by 20%, that’s money in the bank.”

Madnick’s research shows that gaps in corporate culture are often culprits in high-profile hacks, not just the Microsoft example. Prevention, he says, is as much about foresight as hindsight. In a recent article, he cited MIT studies on Equifax and Capital One security breaches of recent years as other prominent examples. “While some risks are true surprises unlikely to be recognized in advance, many are more like the burglar alarm known to be defective,” he said.

Equifax and Capital One did not respond to requests for comment.

Madnick described the corporate mentality as most often “systematic, semi-conscious decision making.” That means management decisions are made without analyzing the cyber risks that are being introduced by the decision. Tying executive compensation to security aims won’t necessarily mean that approach evaporates from a corporate culture, but he said it has symbolic resonance, and from that symbolic register, the practical may indeed follow.

‘An annoyance and a profit center’

For Microsoft, the stakes are higher than for most organizations. Its platforms and systems are so omnipresent — in business and government — that it’s essentially impossible to live without it. “There’s no alternative to Microsoft, from a productivity standpoint. You have to do insane things to try to work without it,” said Ryan Kalember, executive vice president of cybersecurity strategy at cybersecurity vendor Proofpoint.

Adding to the complexity of Microsoft’s unavoidability, he said, is the layered nature of its platforms, in which succeeding iterations are often buttressed by legacy applications stretching back to the 90s, before security threats remotely resembling what now exists.

The U.S. government has called on the largest, and oldest, tech companies to update systems that both businesses and consumers rely on. Last year, Cybersecurity and Infrastructure Security Agency director Jen Easterly said in a CNBC interview that cybersecurity is consumer safety, and compared it to automotive regulations. “Technology companies who for decades have been creating products and software that are fundamentally insecure need to start creating products that are secure by design and secure by default with safety features baked in,” she said. 

Legacy platforms are far easier to plug into and build on rather than deploying a new system entirely, but “it’s a security nightmare,” Kalember said. “One MS365 for everybody from the State Department to Joe’s Crab Shack is a fine business model, it just doesn’t lend itself well to traditional security measures.”

The architectural principles built into some of these legacy systems were designed “when ransomware was really a thing that simply didn’t exist – except on floppy disks,” he said. This has led to the company accruing massive amounts of what is called “technical debt” — decades of it — that can be abused by nation-stated and allow foreign intelligence agencies “to steal anything they want,” he added. 

Microsoft is caught between two competing impulses, with security “a combination of an annoyance and a profit center,” Kalember said. It’s a profit center because Microsoft is the world’s largest cybersecurity vendor, reaching $20 billion in annual revenue last year. That makes the compensation move “a good gesture,” he said, but he added, “without specifics behind it, it’s very difficult to assess.” 

No details on how Microsoft pay will be influenced

The lack of details on the compensation formula makes it impossible to properly evaluate the incentive. Many companies that adopted ESG metrics did so only in the bonus portion of executive pay, not the long-term incentive plan, which is much more significant. “That’s putting your money where your mouth is,” Shah said.

A bonus may comprise, on average, 20% of executive pay, and within the bonus pool specifically, non-core financial metrics such as ESG only contribute 20% of a potential total bonus payout. “When you have 20% of overall [bonus] compensation and divvy it up into a few different metrics, how much are you really tying something like cyber to it?” Shah said.

Long-term incentive plans tied to equity grants, especially in tech, are where the real money is made, and that’s where these types of non-core financial metrics are low in prevalence. That would be the ideal place within a compensation plan to set pay against long-term cybersecurity and corporate goals, but it is difficult for firms to conceive of two-to-three year goals related to cybersecurity, consumer privacy and data breaches that can be measured like sales and profit. “It will be a challenge,” Shah said. “Is it the number of incidents? The caution I have is the same as with ESG: you want to make sure not only the relevance is there, but you also want to make sure there are quantifiable goals. In a rush to adopt, if it’s subjective, then it is less meaningful for shareholders.”

Boards of directors already have the discretion to hold executives accountable each year and decide to do downward adjustments on bonuses, based on performance, including data breaches. To date, this type of bonus incentive/punishment has been mostly limited to chief information security officers, according to Mike Doonan, managing director at SPMB, an executive search firm where he specializes in technology. In his view, it’s an imperfect comparison to look at the history of bonus pay tied to metrics such as worker safety, since many hacks occur due to third-party vulnerabilities, which are often beyond the company’s direct control. But Doonan said he could see this type of executive incentive being adopted more broadly, “because it’s good PR to say security is a top priority across the entire executive suite, and it might result in improvements.” But he thinks there is an even better way to shore up corporate defense: “saving the bonus pool and investing those dollars into security programs.”

Continue Reading

Technology

Amazon plans to give Alexa an AI overhaul — and a monthly subscription price

Published

on

By

Amazon plans to give Alexa an AI overhaul — and a monthly subscription price

Amazon plans to give Alexa an AI overhaul

Amazon is upgrading its decade-old Alexa voice assistant with generative artificial intelligence and plans to charge a monthly subscription fee to offset the cost of the technology, according to people with knowledge of Amazon’s plans. 

The Seattle-based tech and retail giant will launch a more conversational version of Alexa later this year, potentially positioning it to better compete with new generative AI-powered chatbots from companies including Google and OpenAI, according to two sources familiar with the matter, who asked not to be named because the discussions were private. Amazon’s subscription for Alexa will not be included in the $139-per-year Prime offering, and Amazon has not yet nailed down the price point, one source said.

Amazon declined to comment on its plans for Alexa. 

While Amazon wowed consumers with Alexa’s voice-driven tasks in 2014, its capabilities could seem old-fashioned amid recent leaps in artificial intelligence. Last week, OpenAI announced GPT-4o, with the capability for two-way conversations that can go significantly deeper than Alexa. For example, it can translate conversations into different languages in real time. Google launched a similar generative-AI-powered voice feature for Gemini. 

Some interpreted last week’s announcements as a threat to Alexa and Siri, Apple‘s voice assistant feature for iPhones. NYU professor Scott Galloway called the updates the “Alexa and Siri killers” on his recent podcast. Many people use Alexa and Siri for basic tasks, such as setting timers or alarms and announcing the weather.

The development of new AI chatbots in recent months has increased the pressure internally on a division that was once seen as a darling of Amazon founder Jeff Bezos, according to the sources — but has been subject to strict profit imperatives since his departure. 

Three former employees pointed to Bezos’ early obsession with Alexa, describing it as his passion project. Attention from Bezos resulted in more dollars and less pressure to make a return on those funds immediately. 

That changed when Andy Jassy took over as CEO in 2021, according to three sources. Jassy was charged with rightsizing Amazon’s business during the pandemic, and Alexa became less of a priority internally, they said. Jassy has been privately underwhelmed with what modern-day Alexa is capable of, according to one person. The Alexa team worried they had invented an expensive alarm clock, weather machine and way to play Spotify music, one source said.  

For instance, Jassy, an avid sports fan, asked the voice assistant the live score of a recent game, according to a person in the room, and was openly frustrated that Alexa didn’t know an answer that was so easy to find online. 

When reached for comment, Amazon pointed to the company’s annual shareholder letter released last month. In it, Jassy mentioned that the company was building a “substantial number of GenAI applications across every Amazon consumer business,” adding that that included “an even more intelligent and capable Alexa.”

The team is now tasked with turning Alexa into a relevant device that holds up amid the new AI competition, and one that justifies the resources and headcount Amazon has dedicated to it. It has undergone a massive reorganization, with much of the team shifting to the artificial general intelligence, or AGI, team, according to three sources. Others pointed to bloat within Alexa, a team of thousands of employees.

As of 2023, Amazon said it had sold more than 500 million Alexa-enabled devices, giving the company a foothold with consumers. 

Alexa, were you too early?

Apple, Amazon and Google were early movers with their voice assistants, which did employ AI. But the current wave of advanced generative AI enables much more creative, human-sounding interactions. Apple is expected to unveil a more conversational Siri at its annual developers conference in June, according to The New York Times. 

Those who worked on the Alexa team describe it as a great idea that may have been too early, and that it’s going to be hard to turn the ship around. 

There’s also the challenge of finding AI engineering talent, as OpenAI, Microsoft and Google recruit from the same pool of academics and tech talent. Plus, generative AI workloads are expensive thanks to the hardware and computing power required. One source estimated the cost of using generative AI in Alexa at 2 cents per query, and said a $20 price point was floated internally. Another suggested it would need to be in a single-digit dollar amount, which would undercut other subscription offerings. OpenAI’s ChatGPT charges $20 per month for its advanced models. 

Still, they point to Alexa’s installed user base, with devices in hundreds of millions of homes, as an opportunity. Those who worked on Alexa say the fact that it’s already in people’s living rooms and kitchens makes the stakes higher, and mistakes more costly if Alexa doesn’t understand a command or provides unreliable information. 

Amazon has been battling a perception that it’s behind in artificial intelligence. While it offers multiple AI models on Amazon Web Services, it does not have a leading large language model to unseat OpenAI, Google or Meta. Amazon spent $2.75 billion backing AI startup Anthropic, its largest venture investment in the company’s three-decade history. Google also has an Anthropic investment and partnership.

Amazon will use its own large language model, Titan, in the Alexa upgrade, according to a source.  

Bezos is among those who have voiced concern that Amazon is behind in AI, according to two sources familiar with him. Bezos is still “very involved” in Amazon’s AI efforts, CNBC reported last week, and has been sending Amazon executives emails wondering why certain AI startups are picking other cloud providers over AWS. 

Don’t miss these exclusives from CNBC PRO

Jeff Bezos still 'very involved' in Amazon's AI efforts, sources tell CNBC

Continue Reading

Trending