AI Task Costs Are Hard To Tally: Only 11% of Businesses Can Forecast Spend

Not being able to forecast how much it costs to use AI for a specific task is another impediment to faster and broader AI adoption by corporate customers. That unpredictability makes it harder to budget and risks unexpectedly - and unaffordably - high spending.

This does not mean businesses will stop experimenting with and using AI, but it may mean that they are becoming more cautious about how they do so. The problem with that is that AI companies and their investors have been projecting revenues and profits based on maximized expectations. Forecasts of spending accuracy will increase with experience, but that 'in the meantime' period  could take more than impatient investors and lenders are willing to wait. JL

Jillian Vordick and Stephanie Stamm report in the Wall Street Journal:

A recent study found that only 11% of nearly 400 businesses surveyed were able to accurately forecast AI spending. Unlike traditional software, AI behaves more like a human worker: It takes action, makes decisions, sometimes even makes mistakes—all on the clock. While more-advanced models tend to cost more per token, they can sometimes perform tasks more efficiently, leading to lower overall costs. Likewise, asking a “cheap” model to do something it isn’t suited to handle could cause a token run-up.“Total costs depend on many factors for a given task, which can make it hard to forecast new and evolving technology with precision.” As more companies add AI to their daily workflows, they will also need strategies to manage their use

Russian 11th Tank Regiment CO, Who Executed POWs, Killed By Ukraine Strike

The commander of Russia's 11th Tank Regiment was killed by a direct Ukrainian artillery strike on his vehicle as he was inspecting his unit's forward positions. 

If that sounds more like intent and random luck, it probably is. Karaev had been accused of murdering Ukrainian prisoners of war. JL

Vlad Litnarovych reports in United 24:

Major Alexei Karavaev, commander of Russia’s 11th Tank Regiment, was killed by a direct hit from a Ukrainian artillery round on his vehicle while he was traveling to inspect forward positions in the Kharkiv sector. Karavaev is (an accused) war criminal known by the call sign “Father,” who executed prisoners of war in the Kharkiv Region. Investigators linked him to the unit through phone-book records in which his number had reportedly been saved under entries including “Sasha’s Battalion Commander Vovchansk” and “Father SVO 82 Regiment 1 Battalion.” 

The Front Is Still There But Ukraine's Soldiers Are Running A Robot War On Screens

Among the lessons learned during Ukraine's Operation Vivaldi is that there will soon be as many new types of military ground drones as there are soldiers. 

They are successfully preceding and accompanying infantry assaults, not just reducing human risk, but increasing the odds of Ukrainian offensive goal achievement. JL

The World Strategist reports:

A Ukrainian soldier crouches in a dugout in the Donetsk region, eyes on a video screen. Miles away, a tracked robot the size of a wheelbarrow is inching toward a Russian bunker. There is no hope of survival. It carries 30 kilograms of explosive. The robot explodes. The bunker crashes. “Not one Ukrainian foot soldier's life has been put at risk." In assault operations, unmanned aerial vehicles and ground robotic systems always move ahead of infantry. The trenches are still there. But there are no soldiers in them now. They are in bunkers, running robots. Operation Vivaldi has shattered not only Russian defenses but the very assumptions about how wars are fought.

Armored Russian Assault On Huiliapole "Crushed" By Ukrainian Forces

In the 'here we go again' merry-go-round of evolving Russian offensive tactics, infiltration has again been supplanted by mechanized attacks with up-armored vehicles in the Huilaipole sector. 

But Ukraine's drones and their operators have become even more effective, resulting in the destruction of the mechanized assaults before they can reach Ukrainian lines - again. JL

RFU News reports:

At Huiliaipole, in a desperate bid to overturn their stalling progress as infiltrations become less viable, Russian armored assaults return as an option to sustain offensive pressure. They are launching assaults with heavily up-armored mechanized vehicles, frequently accompanied by Russian soldiers driving motorbikes or other all-terrain vehicles. (But) these fast-moving Russian assaulters rarely make it to their mission target alive. Additionally, the Ukrainians have discovered that while extra armor cages protect it against direct strikes, dropped munitions are extremely effective against them. Attacks using heavily up-armored vehicles that only rarely worked the year before are now even less effective through Ukraine’s increased use of bomber drones, FPV drones, and pilots that allow them to take out both the fast-moving all-terrain vehicles and the heavy armor simultaneously.

OpenAI's Safety Leader Quits: "Culture Is Broken, Not Being Nearly Careful Enough"

The spate of recent high level AI company resignations over safety issues are not even mentioning the 'e word' (extinction). They are concerned about the 'parabolic' rise in CVE - common vulnerabilities and exposures. Which sounds pretty anodyne, even boring, as it is meant to do. 

But their frustration and rising anxiety has to do primarily with the culture in the AI and related big tech companies. Some consider it unfailingly optimistic - certainly a Silicon Valley trait - but with an airy dismissal of risk, driven in large measure by intellectual arrogance with a significant dollop of sugar plum wealth visions. Investors do not want to hear about risks, because they reduce speed and cost money. And those on the receiving end of their largess do not want the funds - nor the adulation-  to stop flowing. Humanity will probably look back on these warnings and wish  something had been done, but the history of Silicon Valley suggests that it prefers to learn its lessons the hard way. JL  

Dan Milmo reports in The Guardian:

David Robinson, a safety leader at OpenAI, who led the writing of safety reports that accompanied the ChatGPT developer’s product releases, has quit the company, warning that its culture was broken and that AI firms were not “being nearly careful enough” about developing the technology. OpenAI has “unimpeded optimism” about solving problems as they arose (and) that this internal culture meant safety failures would only grow as systems became more capable. “As the company sprints from one launch to the next, it is failing to achieve the level of care that is needed. Given today’s risks, frontier labs need to run like nuclear power plants or busy airports, with layers of redundancy and careful, time-consuming planning, so that the occasional and inevitable human error does not open a door to disaster.”

Russia's Frontline War Is Going Badly and Getting Worse

The latest battlefield reports from Ukraine paint a consistently negative picture for Russia: territorial losses are rising as are casualties. 

And as if that weren't bad enough, the incidence of Russian troops surrendering, claiming they have been abandoned without food, water, ammunition or major weapons support. To make it even more embarrassing, Russian soldiers are increasingly surrendering not just to Ukrainian soldiers, but to ground drones. JL

Phillups O'Brien reports in his substack:

The front line war is not going well for Russia and seems to be getting worse. New generation, purpose built Unmanned Ground Vehicles (UGVs) are also operating daily. One of the most interesting reports of these was the use of new UGV grenade launchers. These UGVs approached Russian positions, fired at them, then continued toward the Russian shelters before detonating inside. Maybe the most interesting detail from these attacks was that they were able to remotely control these vehicles with their pilots in safe positions “tens of kilometres” behind the lines. Considering the way technological advances and adaptations go in this war, this distance should grow—and that means fewer Ukrainian casualties and more Russian problems

Russia Can Form Only 1 New Division, Down From 5 Two Years Ago

Russia has announced the formation of a new Motor Rifle Division. While some might take this as a sign of strength, the reality is that three such divisions were created in 2025 and seven in 2024. 

That context reveals that the Kremlin's ability to create new military formations has been significantly curtailed, a reality consistent with reports of Russia's inability to recruit more troops due to Ukraine's enhanced ability to inflict casualties on the Kremlin's forces. JL 

Defense Express reports:

The 84th Motor Rifle Division has been formed within Russia's 3rd Combined Arms Army. (But) the emergence of one new division in 2026, compared with three in 2025, represents a relatively low rate of new division formation. By comparison, seven new divisions were reportedly formed in 2024. This came against the backdrop of the russian Army expanding from 12 divisions in 2022 to 23 by early February 2025.