Month: November 2024

Will the CHIPS Act survive a Trump presidency?

November 12, 2024 /

President-elect Donald J. Trump has made no bones about it: legislation passed under the Biden Administration could be on the chopping block once he enters office on Jan. 20, including the bipartisan CHIPS and Science Act.

The CHIPS Act allocates billions of dollars in funding and tax incentives to semiconductor companies to help bring chip manufacturing back to the US.

On the Joe Rogan Experience podcast late last month, Trump said CHIPS Act is “so bad,” and said instead of helping fund new fabrication (fab) and R&D centers, the US should have put tariffs on overseas semiconductor makers. He compared the semiconductor industry to the auto industry as to how tariffs could work to bring back manufacturing.

Trump said, “paying a lot of money to have people build chips, that’s not the way. You tariff it so high that they will come and build their chip companies for nothing. We put up billions of dollars for rich companies to come in and borrow the money and build chip companies here, and they’re not going to give us the good companies anyway. Taiwan, they stole our chip business. They want us to protect [them]. They don’t give us any money to protect them.”

Trump was in all likelihood referring to Taiwan Semiconductor Manufacturing Company (TSMC), the world’s largest producer of computer chips. Computerworld reached out to Trump campaign officials after last week’s election, but did not receive a response.

TSMC is currently building two major semiconductor manufacturing facilities in Arizona — one a 5nm chip fab, the other a 3nm fab plant. The semiconductor designer and manufacturer is being promised $6.6 billion in CHIPS Act funding; in return, the company pledged to bring its most advanced 2nm process technology to US shores and added plans for a third fabrication plant to its Arizona site.

TSMC CEO CC Wei told investors last month he expects volume production of the company’s first Arizona fab to start in early 2025. “We are confident [it will] deliver the same level of manufacturing quality and reliability…as from our fabs in Taiwan,” Wei said. “Our second and third fabs will utilize more advanced technologies based on our customer’s needs. The second fab is scheduled to begin volume production in 2028 and our third fab will begin production by the end of the decade.”

Intel clean room technician

An Intel manufacturing technician at Fab 11X in Rio Ranco, NM, inspects a semiconductor wafer.

Intel Corp.

Jack Gold, principal analyst with tech industry research firm J. Gold Associates, said Trump’s plan to enact tariffs on oversea chip makers is simply “wrong.”

“Tariffs are a penalty, while the CHIPS act is an incentive,” he said. “Incentives almost always work better than penalties. Also, you can put all the tariffs you want on chips, but it still takes three-to-four years to get a new fab up and running, and it’s extremely capital intensive.”

New semiconductor fabrication facilities cost from $20 billion to $40 billion, and many enterprises can’t afford that kind of investment without some form of subsidies or tax break, Gold said.

Additionally, every electronic device today has a chip in it, and so tariffs would likely increase the cost of products from cars and smartphones to toasters.

The CHIPS Act was passed overwhelmingly in 2022 by members of both houses of Congress to address computer chip supply chain shortages that surfaced during the COVID-19 pandemic. The legislation provided the US Department of Commerce (DoC) with $52.7 billion for a suite of programs under the CHIPS for America program to “revitalize” the US position in semiconductor research, development, and manufacturing.

“New fabs are a very high-risk venture, so with the CHIPS Act subsidies, the government is basically saying it will take on part of that risk,” Gold said.

To date, the DoC has allocated, but not dispensed, about $32 billion in funding among chipmakers, including Intel, Samsung, Micron, TSMC, and Texas Instruments, all of whom have unveiled plans for a number of new US chip fabrication plants. In return, those chip designers and makers have pledged about $300 billion in current and future projects in the US, according to the White House.

With the CHIPS Act spurring them on, the likes of Qualcomm, in partnership with GlobalFoundries, also said it would invest $4.2 billion to double chip production in its Malta, NY facility.

In addition to Trump’s opposition, House Speaker Mike Johnson said recently that Republicans will likely repeal the CHIPS Act. Johnson, who voted against the act, later walked his comments back, saying he would like to “streamline” it, according to The Associated Press.

Rep. Brandon Williams (R-NY) said he spoke to Johnson after his remarks were published. “He apologized profusely, saying he misheard the question,” Williams said in a statement. “He clarified his comments on the spot and I trust local media to play his full comments on supporting repatriation of chips manufacturing to America.”

Semiconductor fab in Phoenix

TSMC Arizona’s first fab will produce 5nm semiconductor process technology; it is scheduled to begin production in the first half of 2025. The second fab will utilize a 3nm process and a third plant is expected to use 2nm process technology. Those plants will be in operation beginning in 2028 and beyond.

Shutterstock/Wirestock Creators

“As I have further explained and clarified, I fully support Micron coming to Central NY, and the CHIPS Act is not on the agenda for repeal,” Johnson said in his own statement. “To the contrary, there could be legislation to further streamline and improve the primary purpose of the bill — to eliminate its costly regulations and Green New Deal requirements.”

New York State has become a major hub of semiconductor development. Micron plans to invest $100 billion to build a memory chip fabrication plant in the state that will be the size of 40 football fields and create about 50,000 jobs. It may spend up to an additional $100 billion over the next 20 years on the facility. In addition, the Biden Administration just announced plans to spend about $825 million to create a flagship national semiconductor R&D center in upstate New York, where the government-funded NanoTech Complex already exists.

Micron highlighted the bipartisan support for the CHIPS and Science Act, with a spokesperson saying it’s “a law that represents an important step toward solidifying American semiconductor and technology leadership for decades to come.”

Repealing the legislation, in any event, would take an act of Congress, and the CHIPS Act has strong bipartisan support among senior republicans, including Senate Minority Leader Mitch McConnell. McConnell argued that bolstering US semiconductor production was vital for both national security and economic competitiveness.

Sen. Todd Young (R-IN) was a key CHIPS Act supporter who emphasized the importance of semiconductor manufacturing for both economic growth and national defense and pushed for the bill’s provisions to help American businesses compete globally. Young echoed others in heralding its “broad bipartisan support, and the massive private investments spurred since then have made the legislation even more popular.

“If there are any regulations that can be streamlined to create even more jobs from our growing semiconductor industry in Indiana and across the country, count me in. But I’m confident the CHIPS Act is here to stay,” Young said in a response to Computerworld.

The CHIPS Act has spurred $450 billion in private investment across 28 states, creating 58,000 jobs, according to the Semiconductory Industry Association. 

Despite widespread bipartisan backing, some members of Congress expressed concerns about certain provisions, such as the level of government subsidies or the potential for the bill to benefit only a few large tech companies. Still, the majority of both Democrats and Republicans recognized the strategic importance of boosting semiconductor production on US soil.

A DoC spokesperson pointed to the “overwhelming bipartisan support” for the act’s more than $400 billion in total investments as well as projections it will create more than 125,000 jobs. “Our team continues to implement this bipartisan law in accordance with statute, including announcing more than $36 billion in proposed funding for manufacturing incentives and several key R&D components. We will have more announcements in the coming weeks,” the spokesman said.

TSMC declined comment on the act’s future. Intel, which completed building a new fab in New Mexico and is awaiting CHIPS Act funds for that, was promised a total of $8.5 billion to support investments for fabs, packaging facilities and R&D centers among four states, including Arizona, Oregon, and Ohio.

Ohio is receiving a significant portion of the funding for semiconductor manufacturing. In particular, Intel is investing $20 billion to build two new semiconductor fabrication plants in Licking County, Ohio. Vice President-elect J.D. Vance is currently a US senator in his home state of Ohio.

“I’m betting that lawmakers, and especially Republican lawmakers from certain states that will benefit from the chips act (like Ohio and Arizona, and now New York), will have a lot of push back on [Trump’s] idea,” Gold said.

An Intel spokesperson pointed out that the idea behind the measure began during the first Trump Administration and it continues to maintain strong bipartisan support.“Restoring America’s semiconductor manufacturing leadership is integral to the country’s economic competitiveness and national security,” the spokesperson said. “As the only American company that designs and manufactures leading-edge chips, Intel has a critically important role to play, and we look forward to working with the Trump Administration on this shared priority.”

Over the past 30 years, the US share of global semiconductor production has fallen from 37% to just 12%, according to White House figures. Meanwhile, China’s share of chip manufacturing has grown nearly 50% over the past two years and now comprises about 18% of the world’s supply. That decline in domestic chip production was exposed by a worldwide supply-chain crisis during the Covid-19 pandemic that led to calls for reshoring manufacturing to the US — and ultimately the CHIPS Act.

Over the next three to four years, as new fabrication, packaging and R&D centers are built on US soil, the cost of production of semiconductors is likely to rise by 10% to 20%, according to Gold.

Additionally, it will cost 10% to 20% more to build chips in the US than the Far East, so it will make more sense to focus on higher-end and not commodity chips for US production. Unless Trump’s idea for tariffs is to use the money to help fund new fabrication plants in the US, which is not what the President-elect stated, it would be doing the very thing the CHIPS Act is already accomplishing, Gold said.

US orders TSMC to halt advanced chip exports to China

November 11, 2024 /

In a significant escalation of US efforts to limit China’s access to advanced technology, the Department of Commerce has reportedly mandated Taiwan Semiconductor Manufacturing Co. (TSMC) to cease shipments of high-performance AI chips to Chinese customers.

The directive, effective Monday, restricts the export of TSMC’s 7-nanometer and more advanced processors, which are widely used in AI applications, Reuters reported.

The US Commerce Department’s latest move specifically targets chips that can power AI accelerators and GPUs, with a particular focus on halting indirect access to restricted technology by Chinese companies like Huawei, which the US considers a national security threat.

This directive, marking a new chapter in US-China tech tensions, applies to several key players in China’s AI ecosystem, potentially impacting companies beyond Huawei.

TSMC declined to comment on the said matter citing “market rumor.”

“TSMC is a law-abiding company and we are committed to complying with all applicable rules and regulations, including applicable export controls,” the chip maker said.

A query to the US Commerce Department did not elicit any response.

In another distantly related development, the Taiwanese government has said that the country’s law prevents TSMC from producing its 2nm chips — TSMC’s hitherto most advanced chip —  abroad.

“Since Taiwan has related regulations to protect its own technologies, TSMC cannot produce 2-nanometer chips overseas currently,” Taipei Times said quoting Minister of Economic Affairs J W Kuo.

Kuo made the remarks while addressing concerns that TSMC may have to accelerate 2-nm chip production in its Arizona fabs following Donald Trump’s re-election as US president.

TSMC is the main supplier of chips, including its most advanced one, for Nvidia and Apple and the US largely depends on the Taiwanese firm to further its technological advancements in the AI space.

TSMC’s involvement: The Huawei incident

This stringent order follows a recent finding that a TSMC-manufactured chip had been integrated into Huawei’s Ascend 910B, an advanced AI processor released in 2022.

A teardown analysis by research firm, Tech Insights, revealed the presence of TSMC technology within Huawei’s product, hinting at an export control violation and triggering the US crackdown.

The revelation prompted TSMC to inform the Commerce Department, shedding light on Huawei’s use of intermediaries to potentially bypass US trade restrictions.

The US directive mandates that any advanced product containing over 25% American technology require an export license — a requirement Huawei circumvented by procuring chips indirectly through third parties.

Impact on Chinese tech giants and the semiconductor market

The directive impacts numerous other entities in China’s technology landscape. In addition to Huawei, major AI-driven companies such as Alibaba and Baidu, which design and use similar processors, will face increased scrutiny.

Although the US regards them as competitors to Huawei, the move aims to curb any potential diversion of restricted technology for unauthorized AI applications in China.

Moreover, the order raises questions about TSMC’s ability to navigate US-imposed restrictions while continuing to serve clients in one of its largest markets.

Reports initially suggested that TSMC’s decision to halt chip shipments was voluntary, but it has since become clear that it was a response to direct US government orders.

However, the restriction on AI chips excludes automotive and consumer-grade chips, signaling that China’s AI and defense-related developments are the primary targets.

Growing tensions and US commitment to export control

The US has steadily intensified its stance against the use of American technology by companies that the government deems a security threat. By tightening export controls, the US aims to prevent China from leveraging AI and semiconductor advancements in ways that could counter US interests.

This latest directive follows broader efforts to restrict China’s technological capabilities, underscoring the US commitment to export control enforcement amidst ongoing geopolitical friction.

As the implications of the US directive continue to unfold, TSMC and other semiconductor producers may face a complex path ahead in balancing regulatory compliance with business needs in the Asia-Pacific region.

The FTC’s ‘Click-to-Cancel’ rule for subscriptions is long overdue

November 11, 2024 /

Using the Monarch personal finance program, I went through my finances recently and found that I pay $510 every month on various subscriptions. (That’s not counting things such as my Internet bill, $120 a month for AT&T 2Gbps fiber.) I’m talking about Netflix, Google One, The Wall Street Journal, and other services and publications I actually want. 

But there were also over $100 worth of subscriptions that, frankly, I’d forgotten about  and no longer wanted or needed. That’s real money.

So, how do I get rid of them? Today, I have to dig into every last lousy one of them and jump through numerous hoops to cancel — but that may not be the case for much longer. 

The US Federal Trade Commission (FTC) last month announced a “click-to-cancel” rule aimed at making it easier for you and me to end recurring subscriptions and memberships. The new regulation requires sellers to make canceling services as simple as when you initially signed up for them.

As FTC Commission Chair Lina M. Khan explained: “Too often, businesses make people jump through endless hoops just to cancel a subscription. The FTC’s rule will end these tricks and traps, saving Americans time and money. Nobody should be stuck paying for a service they no longer want.”

Amen, sister!

The new regulations aren’t going to affect just Disney+ subscribers and the like. Businesses that rely on Software-as-a-Service (SaaS) — as either users or providers — are going to be affected as well. 

The new rule, which goes into effect six months after being published in the Federal Register, will have significant implications, for example, for providers like Google One and Microsoft 365. 

Here’s how it’s likely to affect these services.

For example, practices like requiring phone calls or in-person visits to cancel will no longer be allowed. If you think that’s an exaggeration, by the way, you clearly haven’t had a Planet Fitness subscription, which required snail-mail or an in-person visit to close out your membership.

Additionally, SaaS providers must provide clear and conspicuous disclosures about subscription terms: For example, automatic renewal information must be clearly stated and cancellation deadlines by which customers must cancel to avoid charges must also be spelled out. 

Under these regulations, you can no longer automatically resubscribe customers. They must consent before automatic renewals take place. Clearly, businesses that use automatic renewals will have to change how they’ll handle subscription renewals.

If your business gets customers by offering free trials that convert to paid subscriptions, you’ll also need to clearly disclose the trial’s terms, including when the trial ends and what charges will occur. And, of course, canceling after a free trial must be as simple as signing up for the trial.

All of this means, of course, that your company will have to update its terms and conditions. You’re going to have to pay your lawyers (as well as your programmers) to address these new rules. 

On the plus side, while none of this will be cheap, the FTC argues that customers will be happier and more likely resubscribe. And new transparent practices could even lead to  stronger customer relationships.

Not everyone is happy about the new regulations. Business organizations such as the Internet & Television Association (NCTA), the Interactive Advertising Bureau, and the US Chamber of Commerce oppose them. They have three major arguments: that the FTC doesn’t have the legal authority to implement the rules; the change will cost companies money; and they’ll force industries to change current cancellation processes that protect consumers or offer better deals.

In other words, it’s exactly what you’d expect them to say. 

Given the click-to-cancel rule is part of the Biden administration’s efforts to combat “junk fees,” you might think it’s dead as a doornail. Usually, I’d agree. But while Kahn has been a lightning rod for both Democrats and Republicans, she has one ally you probably didn’t expect; Vice President-elect J.D. Vance, who said: “I look at Lina Khan as one of the few people in the Biden administration that I think is doing a pretty good job,”

In addition, overall, the rule appears to be quite popular among consumers and consumer advocates. Let’s get real. People are sick of perpetual subscriptions. Their budgets are tight. Even if the FTC regulation costs companies some coin, it’ll be worth it in the long run.

Mistral’s new tool automatically deletes offending content

November 8, 2024 /

French tech company Mistral AI has launched a new online moderation tool based on the AI ​​model Ministral 8B that can detect and remove offensive or illegal posts automatically. (There is still a risk of some misjudgments, however.)

According to Techcrunch, for example, some studies have shown that posts about people with disabilities can be flagged as “negative” or “toxic” even though that’s not the case.

Initially, Mistral’s new moderation tool will support Arabic, English, French, Italian, Japanese, Chinese, Korean, Portuguese, Russian, Spanish and German, with more languages ​​are on the way later. Mistral in July launched a large language model that can generate longer tranches of code faster than other open-source models.

How to install App Store apps onto SSD drives using macOS Sequoia

November 8, 2024 /

Did you know that Apple’s macOS 15.1 Sequoia now lets you install and use applications acquired from the Mac App Store directly onto an external drive? This enhancement is actually particularly useful if your workflow requires you to handle a space-devouring application.

Here’s what you need to know about it and how it works.

What’s changed?

While anyone who is paying attention should already be impressed by the sheer speed and performance of Apple’s new Macs, that performance also means pro users will push the platform to its limits, banging into any inherent challenges to how Macs work.

One of these challenges is the need to optimize the space you have on your Mac when running larger applications — and given the cost of installing additional space on most Apple hardware, there was demand for a lower-cost way to do just that. The solution comes with macOS Sequoia 15.1.

Wait, is this really new?

So you’ve spotted that many Mac apps (downloaded from outside the App Store) allow users to install and use them on external drives. This is not automatically the case for applications downloaded and installed from the Mac App Store,however — these insist on being hosted on the Mac’s own drive. You have always been able to run most apps and macOS from an external drive, but now you can do the same with App Store apps, including Pro Apple apps.

What are the limitations?

There are some limits to the new feature tweak. 

  • The biggest is that you’ll only be able to install applications larger than 1GB in size, which is great for games and pro apps, less great for users of smaller apps, who may just want to manage storage their own way. We can hope Apple lifts the 1GB restriction eventually.
  • The second limitation is the speed of the external SSD; obviously, the speedier it is, the better the offloaded application will perform.
  • The final — and most inconvenient — limitation is that once it is enabled it is not optional. In the future, you’ll need to install any application of 1GB or more on external storage unless you turn the setting off. 

What do you need?

You need to be running macOS 15.1 and have a suitable connected drive. The drive must also be formatted to APFS. To check that this is so, with the drive connected to your Mac, right-click the drive icon in Finder and select “Get Info.”

How to begin installing Mac apps on external drives

Before you use the feature, you need to open the Mac App Store on your Mac.

  • Go to App Store>Settings in the Menu bar.
  • Check the box beside the “Download and install large apps to a separate disk” item in Settings.
  • When you have enabled that setting, you can select the external drive you want to save your applications to.

After that, when you want to install a large application from the Mac App Store, you will need to ensure the external SSD you want to use is connected to your computer.

How to use a Mac app on an external drive

At the risk of sounding obvious, you do need to connect the drive your application is stored on to your Mac to use the application you have hosted there. It is relatively seamless after that — the app will be visible in your Applications folder, opens with a double click and can be used just like any other app. (One thing it does not do is appear in Launchpad.)

Why does it matter?

Cost is the biggest reason this is important. Additional storage in Macs isn’t cheap; it will cost you an additional $600 to slot 2TB of storage inside the base model MacBook Pro, while a good and speedy external SSD should cost you around two-thirds of that, or less if you’re a little more flexible. That cost increases if you are provisioning multiple seats, so in some cases this feature could help you stretch purchasing budgets a little further. Consumer users can also use this to enable them to better explore and learn about professional applications without needing to worry about having enough space on their Mac.

Please follow me on LinkedIn, Mastodon, or join me in the AppleHolic’s bar & grill group on MeWe.

Microsoft adds Copilot AI features to some non-US M365 consumer plans

November 8, 2024 /

Microsoft is bundling its Copilot generative AI (genAI) assistant with consumer Microsoft 365 subscriptions in several countries, the company announced last week. 

Copilot Pro will be included in Microsoft 365 Personal and Family subscriptions in Australia, Malaysia, New Zealand, Singapore, Taiwan, and Thailand, the company said in a statement first spotted by ZDNet. It means users will gain access to Copilot features in apps such as Word, Excel, and PowerPoint. Designer —Microsoft’s text-to-image app — is also included.

Microsoft will also increase the cost of the subscription — prices will vary in each country — though this will be less than the cost of a separate Copilot Pro subscription. Australian customers, for example, will pay an additional $4 AUD a month for M365 Family subscriptions, and an extra $5 AUD for M365 Personal subscriptions, according to The Verge. In comparison, Copilot Pro costs $33 AUD per user each month. 

Customers will be limited in how much they use Copilot in apps, however, with a credit system in place. Those who want unrestricted access will need to pay for a Copilot Pro subscription. 

Microsoft didn’t say whether it plans to extend the changes to consumer M365 subscriptions in other regions, but it’s possible the move is a trial run for US and European markets.  

In the US, Copilot Pro costs an extra $20 per user per month for M365 Family and Personal customers. 

“I suspect this is just the first step in [Microsoft] bundling Copilot to a larger audience,” said Jack Gold,founder and principal analyst at J. Gold Associates. “The initial countries are probably a trial deployment to see how it goes, what the most common uses are, and how much they can charge. I’ll bet that in the next [one to two] quarters, you’ll see a much wider rollout to many other countries.”

It’s also possible the Copilot bundling in consumer M365 subscriptions could presage a similar move for business customers, though there’s no mention of such a move on the horizon just yet. 

Microsoft charges an extra $30-per-user-a-month fee to businesses for access to Copilot in Microsoft 365. Despite considerable interest in the M365 Copilot, businesses have been slow to rollout the genAI assistant widely across their organizations, in part due to high costs and a perceived lack of value.

It’s likely this will be the case sooner or later: Analysts at Gartner have said they expect genAI features to be included at no extra cost in office software subscriptions by 2028, according to a recent report (subscription required), as vendors seek broader adoption of their AI tools. 

For Microsoft, this could even mean the addition of a new M365 pricing tier — the long-rumored “E7” — that would include premium features currently available as paid-for add-ons, such as Copilot.  

US consumer protection agency bans employee mobile calls amid Chinese hack fears

November 8, 2024 /

The US Consumer Financial Protection Bureau (CFPB) has issued an urgent directive barring employees and contractors from using mobile phones for work-related calls, following a major breach in US telecommunications infrastructure attributed to Chinese-linked hackers.

According to an internal memo, CFPB’s chief information officer advised staff to move sensitive discussions to secure platforms like Microsoft Teams and Cisco WebEx, reported the Wall Street Journal (WSJ).

What if robots learned the same way genAI chatbots do?

November 8, 2024 /

There’s no question that robotics is transforming our world. Thanks to computerized machines, manufacturing, healthcare, agriculture, supply chains, retail, automotive, construction, and other industries are seeing rapidly increasing efficiencies and new capabilities.

One challenge with bringing new robots online is that it’s hard, expensive, and time-consuming to train them for the task at hand. Once you’ve trained them, you have to retrain them with every minor tweak to the system. Robots are capable, but highly inflexible. 

Some of the training is handled by software coding. Other methods use imitation learning, where a person teleoperates a robot (which, during training, essentially functions as a puppet) to kickstart data for robot movement. 

Both approaches are time-consuming and expensive. 

Compounding the difficulty is a lack of standards. Each robot manufacturer uses its own specialized programming language. The interfaces used for teaching robots, especially “teach pendants,” tend to lack the modern attributes of the major, non-proprietary software development environments. (A teach pendant is a handheld control device that enables operators to program and control robots, enabling precise manipulation of the robot’s movements and functions.)

The lack of standards adds both complexity and costs for obvious reasons. Robot programming courses can cost thousands of dollars, and companies often need to train many employees on several robotics programming platforms. 

Because of a lack of standards, because robots are inflexible once trained, and because robot skill development is manual and task-by-task, it is complex, time-intensive, and costly. 

MIT to the rescue?

To solve the enormous problems of robot training, MIT researchers are developing a radical, brilliant new method called Heterogeneous Pretrained Transformers, or HPTs.

The concept is based roughly on the same concept of large language models (LLMs) now driving the generative AI boom. 

LLMs use vast neural networks with billions of parameters to process and generate text based on patterns learned from massive training datasets. 

HPTs work by using a transformer model to process diverse robotic data from multiple sources and modalities. To that data, the model adds and aligns vision and robot-movement inputs in the form of tokens. And all this is processed by an actual LLM. The larger the transformer, the better the robot’s performance. 

While LLMs and HPTs are very different — for starters, every physical robot is mechanically unique and very different from other robots — they both involve vast training datasets from many sources. 

In the case of HPTs, researchers added data from real physical robots and simulation environments and multi-modal data (from vision sensors, robotic arm position encoders, and others). The researchers created a massive dataset for pretraining, including 52 datasets with more than 200,000 robot trajectories.

As a result, HPTs need far fewer task-specific data. And this is early days for the method. As with LLMs, it’s reasonable to expect massive advances in capability with additional data and optimization. 

Researchers found that the HPT method outperformed training from scratch by more than 20% in both simulations and real-world experiments.

Limitations to HPT robot training

While HPTs show promise, they’re still limited and need development. 

Just as even more advanced LLM-based chatbots can “hallucinate” and tend to be polluted with bad data, HPTs need a mechanism for filtering out bad data from the datasets. Nobody wants a powerful industrial robot “hallucinating” and freaking out on the factory floor.

While LLMs and HPTs are similar in concept, LLMs are far more advanced because the available datasets are massively higher. To industrialize the method, the models would need massive quantities of probably simulated data to add to the real-world data. 

As it was during the early days of LLMs, HPT research at MIT is currently averaging below 90% success rates.

According to the researchers, future research should explore several key directions to overcome the limitations of HPT.

To unlock further potential in robotic learning, training objectives beyond supervised learning, such as self-supervised or unsupervised learning, should be investigated. 

It is important to grow the datasets with diverse, high-quality data. This could include teleoperation data, simulations, human videos, and deployed robot data. Researchers need to learn the optimal blend of data types for higher HPT success rates. 

Researchers and later industry will need to create standardized virtual testing grounds to facilitate the comparison of different robot models. (These would likely come from Nvidia.)

Researchers also need to test robots on more complex, real-world tasks. This could involve robots using both hands (bimanual) or moving around (mobile) to complete longer, more intricate jobs. Think of it as giving robots more demanding, more realistic challenges to solve.

Scientists are also looking into how the amount of data, the size of the robot’s “brain” (model), and its performance are connected. Understanding this relationship could help us build better robots more efficiently.

Another exciting area is teaching robots to understand different types of information. This could include 3D maps of their surroundings, touch sensors, and even data from human actions. By combining all these different inputs, robots could learn to understand their environment more like humans do.

All these research ideas aim to create smarter, more versatile robots that can handle a wider range of tasks in the real world. It’s about overcoming the current limitations of robot learning systems and pushing the boundaries of what robots can do.

According to an MIT article on the research, “In the future, the researchers want to study how data diversity could boost the performance of HPT. They also want to enhance HPT so it can process unlabeled data like GPT-4 and other large language models.”

The ultimate goal is a “universal robot brain” that could be downloaded and used without additional training. In essence, HPTs would enable robots to perform far closer to how people act. Specifically, a new, un-trained employee hired to work on an assembly line already knows how to pick things up, walk around, manipulate objects, and identify widgets by sight. They then start out haltingly, gaining confidence with additional skills acquired through practice. MIT researchers see HTP-trained robots as operating the same way. 

This raises obvious concerns about replacing human workers with robots, but that’s a subject for another column. 

In the meantime, I think MIT researchers are onto something here: a new technology that could — and probably will — radically accelerate the industrial robotics revolution. 

Microsoft Ignite 2024 – get the latest news and insights

November 7, 2024 /

Microsoft Ignite 2024 kicks off in Chicago and runs Nov. 19-22.  If you can’t make it to Chicago, no worries. First, the physical event is sold out, according to the Ignite event page. Second, it’s a hybrid event, so you can attend Ignite virtually. 

Whether you’re there physically or online, expect to learn more about the latest technologies from Microsoft — everything from artificial intelligence (AI) to cloud computing, security, productivity tools, and more  In the keynote address, Microsoft CEO Satya Nadella and Microsoft leaders — including Charlie Bell, executive vice president of Microsoft Security and Scott Guthrie, executive vice president of the Microsoft Cloud + AI Group — will share how the company is creating new opportunities across its platforms in this rapidly evolving era of AI.

You can also network with industry experts and Microsoft’s team, IT leaders, and other tech enthusiasts; gain hands-on experience and learn from experts at technical sessions; and learn about new products and services. (Microsoft often announces new products and features at Ignite.)

As you get ready for the event to start, here’s a look back at some of our previous Ignite coverage, as well as recent articles that touch on some of the topics you can expect to see at the event. And remember to check this page often for more on Ignite 2024.

Previous Microsoft Ignite coverage

Microsoft to launch autonomous AI at Ignite

Oct. 21, 2024: Microsoft will let customers build autonomous AI agents that can be configured to perform complex tasks with little or no input from humans. Microsoft announced that tools to build AI agents in Copilot Studio will be available in a public beta that begins at Ignite on Nov. 19, with pre-built agents rolling out to Dynamics 365 apps in the coming months.

Microsoft Ignite 2023: 11 takeaways for CIOs

Nov., 15, 2023: Microsoft’s 2023 Ignite conference might as well be called AIgnite, with over half of the almost 600 sessions featuring AI in some shape or form. Generative AI (genAI), in particular, is at the heart of many of the product announcements Microsoft is making at the event, including new AI capabilities for wrangling large language models (LLMs) in Azure, new additions to the Copilot range of genAI assistants, new hardware, and a new tool to help developers deploy small language models (SLMs) too.

Microsoft partners with Nvidia, Synopsys for genAI services

Nov. 16, 2023: Microsoft has announced that it is partnering with chipmaker Nvidia and chip-designing software provider Synopsys to provide enterprises with foundry services and a new chip-design assistant. The foundry services from Nvidia will be deployed on Microsoft Azure and will combine three of Nvidia’s elements — its foundation models, its NeMo framework, and Nvidia’s DGX Cloud service.

As Microsoft embraces AI, it says sayonara to the metaverse

Feb. 23, 2023: It wasn’t just Mark Zuckerberg who led the metaverse charge by changing Facebook’s name to Meta. Microsoft hyped it as well, notably when CEO Satya Nadella said, “I can’t overstate how much of a breakthrough this is,” in his keynote speech at Microsoft Ignite in 2021. Now, tech companies are much wiser, they tell us. It’s AI at heart of the coming transformation. The metaverse may be yesterday’s news, but it’s not yet dead.

Microsoft Ignite in the rear-view mirror: What we learned

Oct. 17, 2022: Microsoft treated its big Ignite event as more of a marketing presentation than a full-fledged conference, offering up a variety of announcements that affect Windows users, as well as large enterprises and their networks. (The show was a hybrid affair, with a small in-person option and online access for those unable to travel.)

Related  Microsoft coverage

Microsoft’s AI research VP joins OpenAI amid fight for top AI talent

Oct. 15, 2024: Microsoft’s former vice president of genAI research, Sebastien Bubeck, left the company to join OpenAI, the maker of ChatGPT. Bubeck, a 10-year veteran at Microsoft, played a significant role in driving the company’s genAI strategy with a focus on designing more efficient small language models (SLMs) to rival OpenAI’s GPT systems.

Microsoft brings Copilot AI tools to OneDrive

Oct. 9, 2024: Microsoft’s Copilot is now available in OneDrive, part of a wider revamp of the company’s cloud storage platform.  Copilot can now summarize one or more files in OneDrive without needing to open them first; compare the content of selected files across different formats (including Word, PowerPoint, and PDFs); and respond to questions about the contents of files via the chat interface. 

Microsoft wants Copilot to be your new AI best friend

Oct. 09, 2024: Microsoft’s Copilot AI chatbot underwent a transformation last week, morphing into a simplified pastel-toned experience that encourages you…to just chat. “Hey Chris, how’s the human world today?” That’s what I heard after I fired up the Copilot app on Windows 11 and clicked the microphone button, complete with a calming wavey background. Yes, this is the type of banter you get with the new Copilot.

EU launches probe of Corning’s Gorilla Glass for competition violations

November 7, 2024 /

The European Commission has opened a formal investigation into whether US glass producer Corning, known for its Gorilla Glass, might have abused its dominant position in the market for protective glass for electronic devices. Corning’s products are used, among other things, in several of Apple’s and Samsung’s devices.

The Commission suspects the company might have entered into anticompetitive agreements with cell phone makers and glass refiners, including claims for exclusive purchases and discounts based on those pacts. Gorilla Glass has been used in mobile devices for more than a decade.

The agreements might have prevented competitors from entering the market, reducing consumer choice, raising prices and inhibiting innovation. If Corning is found guilty, the company could be fined. Before that happens, Corning will have the chance to respond to the European Commission’s objections and the investigation can be closed if the company fulfills certain commitments.