Project Mariner was a research prototype developed by Google DeepMind that explored human-agent interactions, particularly within web browsers. It automated tasks such as online shopping, information retrieval, and form-filling, aiming to enhance user productivity by delegating routine web-based tasks to an AI agent. Project Mariner operated as an experimental Chrome extension that understands the contents of your screen, including images, code, forms, and more. It could interpret complex goals, plan actionable steps, and navigate websites to carry out tasks, while keeping the user informed and allowing them to intervene at any time. As of May 2025, Project Mariner was available to Google AI Ultra subscribers in the US and was being integrated into the Gemini API and Vertex AI, allowing developers to build applications powered by the agent Google plans to bring Project Mariner’s capabilities to more countries and integrate it into Google Search's AI Mode, which was currently in the Search Labs testing phase. Project Mariner was discontinued on May 4, 2026.
Imo.im
imo.im is a proprietary audio/video calling and instant messaging software service. It allows sending music, video, PDFs and other files, along with various free stickers. It supports encrypted group video and voice calls with up to 20 participants. According to its developer, the service possesses over 200 million users and over 50 million messages per day are sent through it. == History == The product was created as a web-based application in 2005 for accessing multiple chat platforms, including Facebook Messenger, Google Talk, Yahoo! Messenger, and Skype chat. It was developed by Pagebites, which is a subsidiary of Singularity IM, Inc. and required a subscriber's phone number to verify the users' account. In March 2014, support for all third-party messaging networks ended. In January 2018, the app reached 500 million installs. imo.im has implemented end-to-end encryption for its chats and calls, ensuring that the conversations remain private between the sender and receiver.
Token maxxing
Token Maxxing or Token Maxing is a metric used in an attempt to track productivity in the workplace especially for those using Artificial Intelligence (AI) based services. AI services charge for each token which represent units of effort expended by an AI service to solve a problem. Some believe that token consumption equates to productivity and thus can be used as a metric to monitor an employee's work. Supporters believe that higher token usage indicates higher productivity and higher utilization of powerful AI services. This also suggests that those not consuming enough tokens may be less productive and underutilizing powerful AI services. This belief might lead to an environment that incentivizes higher token usage to predict increased productivity. Critics of token maxxing as a metric claim that prudent workers will maximize any metric that management wants increased to gain a workplace advantage. For example: Engineers in the tech industries pressed to consume as many tokens as possible might run several AI agents in tandem, enter longer input prompts, or automate their tasks to maximize their token consumption. To management, this higher token usage may indicate potential productivity, but in reality may cause additional token costs, worker burnout, or actually create more bloated code of lower quality. Another claim is AI service companies potentially benefit from such an emphasis on token consumption and actively encourage the trend. Some developers have publicly advocated the practice. Developer Sigrid Jin, who said he used 50 billion tokens in a single year, has argued that maximizing token consumption is the best way to understand the value of AI, advising others to spend as much on AI usage as they pay in rent to obtain a return on investment. == See Also == Goodhart's law Perverse incentive Jevons Paradox
Highway network
In machine learning, the Highway Network was the first working very deep feedforward neural network with hundreds of layers, much deeper than previous neural networks. It uses skip connections modulated by learned gating mechanisms to regulate information flow, inspired by long short-term memory (LSTM) recurrent neural networks. The advantage of the Highway Network over other deep learning architectures is its ability to overcome or partially prevent the vanishing gradient problem, thus improving its optimization. Gating mechanisms are used to facilitate information flow across the many layers ("information highways"). Highway Networks have found use in text sequence labeling and speech recognition tasks. In 2014, the state of the art was training deep neural networks with 20 to 30 layers. Stacking too many layers led to a steep reduction in training accuracy, known as the "degradation" problem. In 2015, two techniques were developed to train such networks: the Highway Network (published in May), and the residual neural network, or ResNet (December). ResNet behaves like an open-gated Highway Net. == Model == The model has two gates in addition to the H ( W H , x ) {\displaystyle H(W_{H},x)} gate: the transform gate T ( W T , x ) {\displaystyle T(W_{T},x)} and the carry gate C ( W C , x ) {\displaystyle C(W_{C},x)} . The latter two gates are non-linear transfer functions (specifically sigmoid by convention). The function H {\displaystyle H} can be any desired transfer function. The carry gate is defined as: C ( W C , x ) = 1 − T ( W T , x ) {\displaystyle C(W_{C},x)=1-T(W_{T},x)} while the transform gate is just a gate with a sigmoid transfer function. == Structure == The structure of a hidden layer in the Highway Network follows the equation: y = H ( x , W H ) ⋅ T ( x , W T ) + x ⋅ C ( x , W C ) = H ( x , W H ) ⋅ T ( x , W T ) + x ⋅ ( 1 − T ( x , W T ) ) {\displaystyle {\begin{aligned}y=H(x,W_{H})\cdot T(x,W_{T})+x\cdot C(x,W_{C})\\=H(x,W_{H})\cdot T(x,W_{T})+x\cdot (1-T(x,W_{T}))\end{aligned}}} == Related work == Sepp Hochreiter analyzed the vanishing gradient problem in 1991 and attributed to it the reason why deep learning did not work well. To overcome this problem, Long Short-Term Memory (LSTM) recurrent neural networks have residual connections with a weight of 1.0 in every LSTM cell (called the constant error carrousel) to compute y t + 1 = F ( x t ) + x t {\textstyle y_{t+1}=F(x_{t})+x_{t}} . During backpropagation through time, this becomes the residual formula y = F ( x ) + x {\textstyle y=F(x)+x} for feedforward neural networks. This enables training very deep recurrent neural networks with a very long time span t. A later LSTM version published in 2000 modulates the identity LSTM connections by so-called "forget gates" such that their weights are not fixed to 1.0 but can be learned. In experiments, the forget gates were initialized with positive bias weights, thus being opened, addressing the vanishing gradient problem. As long as the forget gates of the 2000 LSTM are open, it behaves like the 1997 LSTM. The Highway Network of May 2015 applies these principles to feedforward neural networks. It was reported to be "the first very deep feedforward network with hundreds of layers". It is like a 2000 LSTM with forget gates unfolded in time, while the later Residual Nets have no equivalent of forget gates and are like the unfolded original 1997 LSTM. If the skip connections in Highway Networks are "without gates," or if their gates are kept open (activation 1.0), they become Residual Networks. The residual connection is a special case of the "short-cut connection" or "skip connection" by Rosenblatt (1961) and Lang & Witbrock (1988) which has the form x ↦ F ( x ) + A x {\displaystyle x\mapsto F(x)+Ax} . Here the randomly initialized weight matrix A does not have to be the identity mapping. Every residual connection is a skip connection, but almost all skip connections are not residual connections. The original Highway Network paper not only introduced the basic principle for very deep feedforward networks, but also included experimental results with 20, 50, and 100 layers networks, and mentioned ongoing experiments with up to 900 layers. Networks with 50 or 100 layers had lower training error than their plain network counterparts, but no lower training error than their 20 layers counterpart (on the MNIST dataset, Figure 1 in ). No improvement on test accuracy was reported with networks deeper than 19 layers (on the CIFAR-10 dataset; Table 1 in ). The ResNet paper, however, provided strong experimental evidence of the benefits of going deeper than 20 layers. It argued that the identity mapping without modulation is crucial and mentioned that modulation in the skip connection can still lead to vanishing signals in forward and backward propagation (Section 3 in ). This is also why the forget gates of the 2000 LSTM were initially opened through positive bias weights: as long as the gates are open, it behaves like the 1997 LSTM. Similarly, a Highway Net whose gates are opened through strongly positive bias weights behaves like a ResNet. The skip connections used in modern neural networks (e.g., Transformers) are dominantly identity mappings.
And–or tree
An and–or tree is a graphical representation of the reduction of problems (or goals) to conjunctions and disjunctions of subproblems (or subgoals). == Example == The and–or tree: represents the search space for solving the problem P, using the goal-reduction methods: P if Q and R P if S Q if T Q if U == Definitions == Given an initial problem P0 and set of problem solving methods of the form: P if P1 and … and Pn the associated and–or tree is a set of labelled nodes such that: The root of the tree is a node labelled by P0. For every node N labelled by a problem or sub-problem P and for every method of the form P if P1 and ... and Pn, there exists a set of children nodes N1, ..., Nn of the node N, such that each node Ni is labelled by Pi. The nodes are conjoined by an arc, to distinguish them from children of N that might be associated with other methods. A node N, labelled by a problem P, is a success node if there is a method of the form P if nothing (i.e., P is a "fact"). The node is a failure node if there is no method for solving P. If all of the children of a node N, conjoined by the same arc, are success nodes, then the node N is also a success node. Otherwise the node is a failure node. == Search strategies == An and–or tree specifies only the search space for solving a problem. Different search strategies for searching the space are possible. These include searching the tree depth-first, breadth-first, or best-first using some measure of desirability of solutions. The search strategy can be sequential, searching or generating one node at a time, or parallel, searching or generating several nodes in parallel. == Relationship with logic programming == The methods used for generating and–or trees are propositional logic programs (without variables). In the case of logic programs containing variables, the solutions of conjoint sub-problems must be compatible. Subject to this complication, sequential and parallel search strategies for and–or trees provide a computational model for executing logic programs. == Relationship with two-player games == And–or trees can also be used to represent the search spaces for two-person games. The root node of such a tree represents the problem of one of the players winning the game, starting from the initial state of the game. Given a node N, labelled by the problem P of the player winning the game from a particular state of play, there exists a single set of conjoint children nodes, corresponding to all of the opponents responding moves. For each of these children nodes, there exists a set of non-conjoint children nodes, corresponding to all of the player's defending moves. For solving game trees with proof-number search family of algorithms, game trees are to be mapped to and–or trees. MAX-nodes (i.e. maximizing player to move) are represented as OR nodes, MIN-nodes map to AND nodes. The mapping is possible, when the search is done with only a binary goal, which usually is "player to move wins the game".
Score bug
A score bug is a digital on-screen graphic which is displayed in a broadcast of a sporting event, displaying the current score and other statistics. It is similar in function to a scoreboard, and is usually placed at either the top or lower third of the television screen. == History == The concept of a persistent score bug was devised by Sky Sports head David Hill, who was dissatisfied over having to wait to see what the score was after tuning into a football match in-progress. The score bug was introduced when Sky launched its coverage of the then newly-formed English Premier League in August 1992. Hill's boss repeatedly demanded that the graphic be removed, describing it as the "stupidest thing [he] had ever seen". Hill defied the boss's demands and kept the graphic in place. ITV introduced a score bug at the start of the 1993–94 football season, and the BBC introduced a score bug towards the end of 1993. The concept was introduced to the United States by ABC Sports and ESPN during coverage of the 1994 FIFA World Cup. Their justification for the graphic was to provide a location for a rotating series of sponsor logos, in order to allow matches to air without commercial interruption. With the acquisition of rights to the National Football League (NFL) by BSkyB's American sibling Fox (a fellow venture of Rupert Murdoch), Hill became the first president of Fox Sports. Under Hill's leadership, Fox introduced a version of the score bug branded as the "Fox Box", which was part of its inaugural season of NFL coverage in 1994. Variety criticized it as an "annoying see-through clock and score graphic" and expressed concern for people "who actually watched the beginning of the game and would rather have their screen clear of graphics". Hill even received a death threat from an irate viewer, with a specific emphasis on him being a "foreigner", but the score bug soon became a ubiquitous feature for American football broadcasts, along with almost all American sports broadcasts in the years that followed. Dick Ebersol of NBC Sports initially opposed the idea of a score bug, as he thought that fans would dislike seeing more graphics on the screen and would change the channel from blowout games if the score was constantly being displayed. Since the 2010s, the on-air design and positioning of some score bugs have been influenced by the needs of Internet video (especially when viewing an event on devices with smaller screens), including bugs noticeably larger than prior iterations designed with television viewing in mind, or designs primarily kept towards the bottom-center of the screen (easing the ability for the bug to remain visible when highlights are cropped for square videos posted on social media). == Details == Score bugs used in team sports typically include the names of both teams, an abbreviation of the team's name, and/or the team's logo; for individual sports, they include the names of individual competitors. In sports where a game clock or playing periods are used, those are generally also displayed as part of the score bug. Some broadcasts also include teams' win-loss records. In 2024, ESPN experimented with adding a persistent win probability meter to its bug in Major League Baseball, which was based on input from its statisticians. === Variations === In addition to the above information, score bugs in some sports include additional information: In baseball, score bugs display the current inning, number of outs, the pitch clock if applicable, and a graphic displaying which bases are occupied; and usually include names of the current pitcher and batter, the pitcher's pitch count, and the number of balls and strikes accrued by the batter. In basketball, score bugs generally include the shot clock, the number of fouls accrued by each team, and whether a team is in the bonus. In cricket, score bugs often take the form of larger dashboards across the bottom of the screen, displaying the current team up and their number of runs, wickets, and overs, a display showing the runs scored and number of balls faced by the current batting partnership, and statistics for the opposing team's bowler (including the number of wickets scored and runs given up). In American football, score bugs usually include the play clock and the down and distance of the current play; they also incorporate graphics indicating when a penalty flag has been thrown. In ice hockey, score bugs display when a penalty or power play is in effect, and often include the number of shots on goal accrued by each team. In golf, Fox popularized the display of a persistent leaderboard graphic in the bottom-right of the screen, usually displaying the top 5. ==== Racing ==== Telecasts of automobile races often include a score bug with the current positions of participants, statistics such as distance behind the leader, and the remaining distance or number of laps. In the mid-2010s, NASCAR broadcasters such as Fox began to transition from horizontal tickers to vertical leaderboards (also referred to as "pylons", in reference to the physical scoring pylons at). The CW differentiated itself by using a horizontal display that divides the field into multiple columns along the bottom of the screen.
Model compression
Model compression is a machine learning technique for reducing the size of trained models. Large models can achieve high accuracy, but often at the cost of significant resource requirements. Compression techniques aim to compress models without significant performance reduction. Smaller models require less storage space, and consume less memory and compute during inference. Compressed models enable deployment on resource-constrained devices such as smartphones, embedded systems, edge computing devices, and consumer electronics computers. Efficient inference is also valuable for large corporations that serve large model inference over an API, allowing them to reduce computational costs and improve response times for users. Model compression is not to be confused with knowledge distillation, in which a smaller "student" model is trained to imitate the input-output behavior of a larger "teacher" model (as opposed to using the "teacher"'s trained parameters or the "teacher"'s training targets). == Techniques == Several techniques are employed for model compression. === Pruning === Pruning sparsifies a large model by setting some parameters to exactly zero. This effectively reduces the number of parameters. This allows the use of sparse matrix operations, which are faster than dense matrix operations. Pruning criteria can be based on magnitudes of parameters, the statistical pattern of neural activations, Hessian values, etc. === Quantization === Quantization reduces the numerical precision of weights and activations. For example, instead of storing weights as 32-bit floating-point numbers, they can be represented using 8-bit integers. Low-precision parameters take up less space, and takes less compute to perform arithmetic with. It is also possible to quantize some parameters more aggressively than others, so for example, a less important parameter can have 8-bit precision while another, more important parameter, can have 16-bit precision. Inference with such models requires mixed-precision arithmetic. Quantized models can also be used during training (rather than after training). PyTorch implements automatic mixed-precision (AMP), which performs autocasting, gradient scaling, and loss scaling. === Low-rank factorization === Weight matrices can be approximated by low-rank matrices. Let W {\displaystyle W} be a weight matrix of shape m × n {\displaystyle m\times n} . A low-rank approximation is W ≈ U V T {\displaystyle W\approx UV^{T}} , where U {\displaystyle U} and V {\displaystyle V} are matrices of shapes m × k , n × k {\displaystyle m\times k,n\times k} . When k {\displaystyle k} is small, this both reduces the number of parameters needed to represent W {\displaystyle W} approximately, and accelerates matrix multiplication by W {\displaystyle W} . Low-rank approximations can be found by singular value decomposition (SVD). The choice of rank for each weight matrix is a hyperparameter, and jointly optimized as a mixed discrete-continuous optimization problem. The rank of weight matrices may also be pruned after training, taking into account the effect of activation functions like ReLU on the implicit rank of the weight matrices. == Training == Model compression may be decoupled from training, that is, a model is first trained without regard for how it might be compressed, then it is compressed. However, it may also be combined with training. The "train big, then compress" method trains a large model for a small number of training steps (less than it would be if it were trained to convergence), then heavily compress the model. It is found that at the same compute budget, this method results in a better model than lightly compressed, small models. In Deep Compression, the compression has three steps. First loop (pruning): prune all weights lower than a threshold, then finetune the network, then prune again, etc. Second loop (quantization): cluster weights, then enforce weight sharing among all weights in each cluster, then finetune the network, then cluster again, etc. Third step: Use Huffman coding to losslessly compress the model. The SqueezeNet paper reported that Deep Compression achieved a compression ratio of 35 on AlexNet, and a ratio of ~10 on SqueezeNets.